A Feasibility Study on Monitoring Residual Sugar and Alcohol Strength in Kiwi Wine Fermentation Using a Fiber‐Optic FT‐NIR Spectrometry and PLS Regression

This work aims to investigate the potential of fiber‐optic Fourier transform‐near‐infrared (FT‐NIR) spectrometry associated with chemometric analysis, which will be applied to monitor time‐related changes in residual sugar and alcohol strength during kiwi wine fermentation. NIR calibration models for residual sugar and alcohol strength during kiwi wine fermentation were established on the FT‐NIR spectra of 98 samples scanned in a fiber‐optic FT‐NIR spectrometer, and partial least squares regression method. The results showed that R² and root mean square error of cross‐validation could achieve 0.982 and 3.81 g/L for residual sugar, and 0.984 and 0.34% for alcohol strength, respectively. Furthermore, crucial process information on kiwi must and wine fermentations provided by fiber‐optic FT‐NIR spectrometry was found to agree with those obtained from traditional chemical methods, and therefore this fiber‐optic FT‐NIR spectrometry can be applied as an effective and suitable alternative for analyses and monitoring of those processes. The overall results suggested that fiber‐optic FT‐NIR spectrometry is a promising tool for monitoring and controlling the kiwi wine fermentation process.     

Preliminary study on the use of near‐infrared reflectance spectroscopy to assess nitrogen content of undried wheat plants

Near‐infrared reflectance (NIR) spectroscopy combined with chemometrics was used to assess nitrogen (N) and dry matter content (DM) and chlorophyll in whole‐wheat plant (Triticum aestivum L). Whole‐wheat plant samples (n = 245) were analysed by reference method and by visible and NIR spectroscopy, in fresh (n = 182) and dry (n = 63) presentations, respectively. Calibration equations were developed using partial least squares (PLS) and validated using full cross‐validation (leave‐one‐out method). Coefficient of determination in calibration (R²_CAL) and the standard error of cross‐validation (SECV) for N content in fresh sample presentation, after second derivative, were 0.89 (SECV: 0.64%), 0.86 (SECV: 0.66%) and 0.82 (SECV: 0.74%) using the visible + NIR, NIR and visible wavelength regions, respectively. Dry sample presentation gave better R²_CAL and SECV for N compared with fresh presentation (R²_CAL > 0.90, SECV < 0.20%) using visible + NIR. The results demonstrated that NIR is a suitable method to assess N concentration in wheat plant using fresh samples (unground and undried). Copyright © 2006 Society of Chemical Industry     

Rapid Estimation of Potato Tuber Quality by Near‐Infrared Spectroscopy

Starch and protein parameters of potato tubers were estimated by near‐infrared (NIR) spectroscopy measurements of the crude potato mash. Calibration was carried out with a condensed data set of original 1481 individual samples from a varying number of varieties and breeding lines, grown at eight locations over a three‐years period. Validation of the models was performed with an independent data set (n = 133). Starch content of potato tubers was determined with the official under‐water weighting procedure of the EU, and could be predicted with 90% confidence. The NIR model of phosphorus content of starch had a prediction confidence of 53%. Total protein content could be predicted, too (62% confidence), whereas the amount of coagulable protein was not predictable (R² = 0.25). Despite the different qualification levels of the models, guiding to concrete prediction tools or to a very rough estimation graduating from “low” to “high”, the NIR technique enables potato starch processing plants to optimise both potato starch quality and processing efficiency.     

Direct and Simultaneous Quantification of Tannin Mean Degree of Polymerization and Percentage of Galloylation in Grape Seeds Using Diffuse Reflectance Fourier Transform‐Infrared Spectroscopy

The direct and simultaneous quantitative determination of the mean degree of polymerization (mDP) and the degree of galloylation (%G) in grape seeds were quantified using diffuse reflectance infrared Fourier transform spectroscopy and partial least squares (PLS). The results were compared with those obtained using the conventional analysis employing phloroglucinolysis as pretreatment followed by high performance liquid chromatography‐UV and mass spectrometry detection. Infrared spectra were recorded in solid state samples after freeze drying. The 2nd derivative of the 1832 to 1416 and 918 to 739 cm^?1 spectral regions for the quantification of mDP, the 2nd derivative of the 1813 to 607 cm^?1 spectral region for the degree of %G determination and PLS regression were used. The determination coefficients (R²) of mDP and %G were 0.99 and 0.98, respectively. The corresponding values of the root‐mean‐square error of calibration were found 0.506 and 0.692, the root‐mean‐square error of cross validation 0.811 and 0.921, and the root‐mean‐square error of prediction 0.612 and 0.801. The proposed method in comparison with the conventional method is simpler, less time consuming, more economical, and requires reduced quantities of chemical reagents and fewer sample pretreatment steps. It could be a starting point for the design of more specific models according to the requirements of the wineries.     

Development of a Single Kernel NIR Barley Protein Calibration and Assessment of Variation in Protein on Grain Quality

In this paper, we describe results from a preliminary experiment on the development of a near infrared reflectance (NIR) calibration for single kernel (SK) % protein content in barley. The SKNIR calibration was developed using kernels from breeding lines and commercial barley varieties with a range in protein content of 7.3% to 16.6% “as is”. The calibration model produced an R² = 0.903, while the validation set had a R² = 0.837 with a standard error of cross validation and a standard error of prediction of 0.8% for both the calibration and validation sets respectively. The calibration was then used to estimate the variation in % protein of 4,000 single kernels from a commercial variety (Gairdner at 9.3% protein) by segregating kernels into six sub‐groups (<7.8%, 7.9–8.3%, 8.4–9.0%, 9.1–9.7%, 9.8–10.4%, >10.5% “as is”). These sub‐groups then had additional grain quality tests carried out including grain size, thousand kernel weight and NIR estimates of % protein, starch, hardness and barley hot water extract (HWE). The results showed an increase in grain size, and a decrease in HWE from the low to high % protein sub‐groups. While only a single variety was used in the SKNIR protein segregation study, the results suggested SKNIR could be used to screen for the variation in grain quality traits based on variation in protein content.     

Quality Evaluation of Shelled and Unshelled Macadamia Nuts by Means of Near‐Infrared Spectroscopy (NIR)

The quality of shelled and unshelled macadamia nuts was assessed by means of Fourier transformed near‐infrared (FT‐NIR) spectroscopy. Shelled macadamia nuts were sorted as sound nuts; nuts infected by Ecdytolopha aurantiana and Leucopteara coffeella; and cracked nuts caused by germination. Unshelled nuts were sorted as intact nuts (<10% half nuts, 2014); half nuts (March, 2013; November, 2013); and crushed nuts (2014). Peroxide value (PV) and acidity index (AI) were determined according to AOAC. PCA‐LDA shelled macadamia nuts classification resulted in 93.2% accurate classification. PLS PV prediction model resulted in a square error of prediction (SEP) of 3.45 meq/kg, and a prediction coefficient determination value (R_p²) of 0.72. The AI PLS prediction model was better (SEP = 0.14%, R_p² = 0.80). Although adequate classification was possible (93.2%), shelled nuts must not contain live insects, therefore the classification accuracy was not satisfactory. FT‐NIR spectroscopy can be successfully used to predict PV and AI in unshelled macadamia nuts, though.     

Multivariate determination of free fatty acids and moisture in fish oils by partial least-squares regression and near-infrared spectroscopy

The oxidative and hydrolytic degradation of lipids in fish oil was monitored using partial least-squares (PLS) regression and near-infrared reflectance (NIR) spectroscopy. One hundred and sixty (n=160) fish oil samples from a fishmeal factory were scanned in transflectance by an NIR monochromator instrument (1100–2500 nm). Calibration models were performed for free fatty acids (FFA), moisture (M), peroxide value (PV) and anisidine value (AV). Coefficients of determination in calibration (R²) and standard errors of cross validation (SECV) were 0.96 (SECV: 0.59) and 0.94 (SECV: 0.03) for FFA and M in g/kg, respectively. The accuracy of the NIR calibration models were tested using a validation set, yielding coefficients of correlation (r) and standard errors of prediction (SEP) of 0.98 (SEP: 0.50) and 0.80 (SEP: 0.05) for FFA and M in g/kg, respectively. Poor accuracy (R²<0.80) was obtained for the NIR calibration models developed for PV and AV. The paper demonstrates that fish oil hydrolytic degradation of lipids, which seriously affect oil use and storage under industrial conditions, can be successfully monitored using PLS regression and NIR spectroscopy by the fishmeal industry.     

Texture Quality Analysis of Rainbow Trout Using Hyperspectral Imaging Method

In this study a hyperspectral imaging system (short wave infrared range from 1000 to 2500 nm) was used to model fish texture by experimental compression test. Partial least square–discriminate analysis modeling technique was used for classifying the samples by linking the hyperspectral information and their measured texture. The R² of cross validation and prediction were 0.97 and 0.96, respectively. The root mean squared errors for cross validation and prediction were 0.07 and 0.09, respectively. Sensitivity and specificity for both class I and II were 1.00. Results indicated that hyperspectral imaging in short wave infrared range has ability to detect texture stiffness of rainbow trouts which is affected by freshness.     

RAPID HIGH PERFORMANCE LIQUID CHROMATOGRAPHIC DETECTION OF FUROSINE (ε‐N‐2‐FUROYLMETHYL‐L‐LYSINE) IN YOGURT AND CHEESE MARKETED IN TURKEY*

Furosine (ε‐N‐2‐furoylmethyl‐L‐lysine) content determination in the yogurt and different cheese types (pickled white, kasar, processed, canned tulum, blue‐veined and mozzarella cheeses) marketed in Turkey was performed using ion‐pair reversed‐phase high performance liquid chromatography (RP‐HPLC). Calibration study (R² = 0.9999), analytical method validation and recovery studies gave satisfactory results. The lowest furosine values were observed in pickled white cheeses (5.35 ± 0.01 to 7.28 ± 0.02 mg/100 g protein). All cheeses except pickled white showed furosine values between 182.16 ± 0.12 (canned tulum) and 261.32 ± 0.10 mg/100 g protein (ripened kasar). The highest content of furosine was observed in whole yogurt (316.47 ± 0.17 mg/100 g protein) which could be because of severe heat treatment and the addition of milk powder during the manufacturing process. The method provides a rapid, reproducible and accurate determination of this Amadori compound (ε‐deoxy‐fructosyl‐lysine) in yogurt and cheese samples.     

Discrimination and classification of adulterants in maple syrup with the use of infrared spectroscopic techniques

Food adulteration is a profit‐making business for some unscrupulous manufacturers. Maple syrup is a soft target of adulterators owing to its simplicity of chemical composition. In this study the use of Fourier transform infrared (FTIR) spectroscopy and near‐infrared (NIR) spectroscopy to detect adulterants such as cane and beet invert syrups as well as cane and beet sugar solutions in maple syrup was investigated. The FTIR spectrum of adulterated samples was characterised and the regions 800–1200 cm^?1 (carbohydrates) and 1200–1800 and 2800–3200 cm^?1 (carbohydrates, carboxylic acids and amino acids) were used for detection. The region between 1100 and 1660 nm in the NIR spectrum was used for analysis. Linear discriminant analysis (LDA) and canonical variate analysis (CVA) were used for discriminant analysis, while partial least squares (PLS) and principal component regression (PCR) were used for quantitative analysis. FTIR was more accurate in predicting adulteration using two different regions (R² > 0.93 and >0.98) compared with NIR (R² > 0.93). Classification and quantification of adulterants in maple syrup show that NIR and FTIR can be used for detecting adulterants such as pure beet and cane sugar solutions, but FTIR was superior to NIR in detecting invert syrups. © 2003 Society of Chemical Industry     

Design of an Optimum Computer Vision‐Based Automatic Abalone (Haliotis discus hannai) Grading Algorithm

An automatic abalone grading algorithm that estimates abalone weights on the basis of computer vision using 2D images is developed and tested. The algorithm overcomes the problems experienced by conventional abalone grading methods that utilize manual sorting and mechanical automatic grading. To design an optimal algorithm, a regression formula and R² value were investigated by performing a regression analysis for each of total length, body width, thickness, view area, and actual volume against abalone weights. The R² value between the actual volume and abalone weight was 0.999, showing a relatively high correlation. As a result, to easily estimate the actual volumes of abalones based on computer vision, the volumes were calculated under the assumption that abalone shapes are half‐oblate ellipsoids, and a regression formula was derived to estimate the volumes of abalones through linear regression analysis between the calculated and actual volumes. The final automatic abalone grading algorithm is designed using the abalone volume estimation regression formula derived from test results, and the actual volumes and abalone weights regression formula. In the range of abalones weighting from 16.51 to 128.01 g, the results of evaluation of the performance of algorithm via cross‐validation indicate root mean square and worst‐case prediction errors of are 2.8 and ±8 g, respectively.     

Visible/Near-Infrared Spectra for Linear and Nonlinear Calibrations: A Case to Predict Soluble Solids Contents and pH Value in Peach

Two sensitive wavelength (SWs) selection methods combined with visible/near-infrared (Vis/NIR) spectroscopy were investigated to determine the soluble solids content (SSC) and pH value in peaches, including latent variables analysis (LVA) and independent component analysis (ICA). A total of 100 samples were prepared for the calibration (n = 70) and prediction (n = 30) sets. Calibration models using SWs selected by LVA and ICA were developed, including linear regression of partial least squares (PLS) analysis and nonlinear regression of least squares-support vector machine (LS-SVM). In the nonlinear models, four SWs selected by ICA achieved the optimal ICA-LS-SVM model compared with LV-LS-SVM and both of them better than linear model of PLS. The correlation coefficients (r _p and r _cv), root mean square error of cross validation, root mean square error of prediction, and bias by ICA-LS-SVM were 0.9537, 0.9485, 0.4231, 0.4155, and 0.0167 for SSC and 0.9638, 0.9657, 0.0472, 0.0497, and −0.0082 for pH value, respectively. The overall results indicated that ICA was a powerful way for the selection of SWs, and Vis/NIR spectroscopy incorporated to ICA-LS-SVM was successful for the accurate determination of SSC and pH value in peach.     

PREDICTION OF THE TEXTURE OF COOKED POULTRY PECTORALIS MAJOR MUSCLES BY NEAR-INFRARED REFLECTANCE ANALYSIS OF RAW MEAT

Near‐infrared reflectance (NIR) spectra were collected on pectoralis major muscles from 90 broiler carcasses ( from 2 different processing sets) to assess the relationship between the NIR and a cutting‐shear instrumental texture test. For the instrumental razor blade test, two instrumental parameters (maximum shearing force and total shear energy) were calculated. Calibration (R²) and validation () coefficients of determination were obtained for predicting the instrumental measurements using the reflectance and its first and second derivatives. Models obtained with the second derivative were adequate when the two groups of samples were analyzed separately. The R² values ranged from 0.90 to 0.95 and from 0.84 to 0.89 for both maximum shear force and total shear energy. The regressions performed on the two sample sets combined did not yield model statistics that were as satisfactory (R² = 0.85–0.86 and = 0.78–0.77), suggesting that a prediction model accurately predicting poultry breast meat tenderness will need a larger and more varied sample set. The results suggest that NIR could be used to predict poultry meat texture and to classify muscles according to tenderness levels.     

A Fourier Transform Infrared Spectroscopy Method for Analysis of Palm Oil Adulterated with Lard in Pre-Fried French fries

Fourier transform infrared spectroscopy with attenuated total reflectance accessory was used to detect the presence of lard in French fries pre-fried in palm oil adulterated with lard. A Fourier transform infrared calibration model was obtained using partial least squares for prediction of lard in a blend mixture of lard and palm oil. The coefficient of determination (R²) of 0.9791 was obtained with 0.5% of detection limit. The error in calibration expressed with root mean square error of calibration was 0.979%. In addition, the error obtained during cross validation was 2.45%. A discriminant analysis test was able to distinguish between fries samples adulterated with lard and samples, which were pre-fried with palm oils. Fourier transform infrared spectroscopy is a fast and powerful technique for quantification of lard present in French fries.      

Quantification of soluble solids in reconstituted açaí (Euterpe oleracea Mart.) pulp using near‐infrared spectroscopy

Açaí consumption is increasing worldwide because of the growing recognition of its nutritional and therapeutic properties. This product is classified based on its soluble solids content (SS), but the determination of SS in pulp is time consuming, tedious and not suitable for modern food processing plants. As near‐infrared (NIR) systems have been implemented to measure various quality attributes of food products, the objective of this study was to evaluate the feasibility of NIR diffuse reflectance spectroscopy to quantify the SS content of açaí pulp. Partial least squares (PLS) regression models were constructed to predict the SS. An optimum PLS model required one latent variable [principal component (PC)1 = 97%] with a root‐mean‐square error of calibration (RMSEC) of 1.06% for the calibration data set and the root‐mean‐square error of prediction (RMSEP) of 1.03% for internal cross‐validation. External validation using an independent data set showed good performance (RMSEP = 1.33% and Rp² = 0.82). NIR spectroscopy is a reliable method with which to determine SS in açaí pulp and thereby to classify açaí pulp according to established minimum quality standards.     

Fourier Transform Infrared Spectroscopy Combined with Multivariate Calibrations for the Authentication of Avocado Oil

Avocado oil is one of the functional oils having high quality and high price in the market. This oil shows many benefits for the human health and is applied in many cosmetic products. The authentication of avocado oil becomes very important due to the possible adulteration of avocado oil with other lower priced oils, such as palm oil and canola oil. In this study, Fourier transform infrared spectroscopy using attenuated total reflectance in combination with chemometrics techniques of partial least squares and principal component regression is implemented to construct the quantification and classification models of palm oil and canola oil in avocado oil. Partial least squares at the wavenumbers region of 1260–900 cm^–1 revealed the best calibration models, having the highest coefficient of determination (R² = 0.999) and the lowest root mean square error of calibration, 0.80%, and comparatively low root mean square error of prediction, 0.79%, for analysis of avocado oil in the mixture with palm oil. Meanwhile, the highest R², root mean square error of calibration, and root mean square error of prediction values obtained for avocado oil in the mixture with canola oil at frequency region of 3025–2850 and 1260–900 cm^–1 were 0.9995, 0.83, and 0.64%, respectively.     

Determination of fat in live farmed Atlantic salmon using non‐invasive NIR techniques

Non‐destructive near‐infrared (NIR) measurements were performed on 100 live, anaesthetised farmed Atlantic salmon, whole weight 1–11 kg, using two different NIR instruments: a grating monochromator instrument equipped with a fibre optic interactance probe, and a diode array instrument measuring diffuse reflectance in a non‐contact mode. Crude fat content was determined using partial least squares (PLS) regression. Full cross‐validation was used to evaluate the performance of the calibration models, expressed as the root mean square error of prediction (RMSEP). For the fibre optic instrument the wavelength range from 800 to 1098 nm resulted in a correlation coefficient of 0.90 and an RMSEP equal to 14 g kg^?1 fat. The diode array instrument using the wavelength range from 900 to 1700 nm gave results of the same accuracy. The measurement times were 21 and 3 s respectively. It is concluded that either instrument could be used to determine the crude fat content in live Atlantic salmon, with good accuracy. © 2003 Society of Chemical Industry     

Non-destructive determination of fat,moisture and protein in salmon fillets by use of near-infrared diffuse spectroscopy

Near infrared (NIR) diffuse spectroscopy was used to determine the fat, moisture and protein contents in whole and ground farmed atlantic salmon fillets. A remote fibre-optic probe was used for NIR measurements on 50 whole salmon fillets. The constituent ranges were: 91-205 g kg^?1 fat, 599-709g kg^?1 moisture and 186-209 g kg^?1 protein. Principal component regression resulted in the following prediction errors for ground salmon fillets, expressed as root mean square error of cross validation: 6.6 g kg-1 fat, 3.8 g kg^?1 moisture and 2.0 g kg^?1 protein. The corresponding prediction errors for non-destructive measurements on whole salmon fillets were 10.8 g kg^?1 fat, 8.5 g kg^?1 moisture and 3.7 g kg^?1 protein. Regression models using the 760-1100 m range gave lower prediction errors than models using the 1100-2500 mm or 760-2500 nm ranges. The results show that fibre-optic probe NIR instruments are suited to determine fat and moisture in whole salmon fillets non-destructively.     

FTIR-ATR Spectroscopy Based Metabolite Fingerprinting as A Direct Determination of Butter Adulterated With Lard

Adulteration of butter with cheaper animal fats, such as lard, has become an issue in recent years. A simple and rapid analytical method of attenuated total reflectance in Fourier transform infrared spectroscopy was developed in order to determine the lard content in butter. The multivariate calibration of partial least square model for the prediction of adulterant was developed for quantitative measurement. The model yielded the highest regression with the correlation coefficient (R²) = 0.999, its lowest root mean square error estimation = 0.0947, and its root mean square error prediction = 0.0687, respectively. Cross validation testing evaluates the predictive power of the model. Partial least square model to be effective as their intercept of R²Y and Q²Y were 0.08 and –0.34, respectively.     

Shortwave-near infrared spectroscopy for determination of reducing sugar content during grape ripening,winemaking, and aging of white and red wines

The feasibility of near infrared (NIR) spectroscopy for predicting reducing sugar content during grape ripening, winemaking, and aging was assessed. NIR calibration models were developed using a set of 146 samples scanned in a quartz flow cell with a 50 mm path length in the NIR region (800–1050 nm), in a fiber spectrometer system working in transmission mode. Principal component analysis (PCA), partial least squares (PLS), and multiple linear (MLR) regressions were used to interpret spectra and to develop calibrations for reducing sugar content in grape, must, and wine. The PLS model based on the full spectral range (800–1050 nm), yielded a determination coefficient (r²) of 0.98, a standard error of cross validation (SECV) of 13.62 g/l and a root mean square error of cross validation (RMSECV) of 13.58 g/l. The mathematical model was tested with independent validation samples (n = 48); the resulting values for r², the standard error of prediction (SEP) and the root mean square error of prediction (RMSEP) for the same parameter were 0.98, 10.84, and 12.20 g/l, respectively. The loading weights of latent variables from the PLS model were used to identify sensitive wavelengths. To assess their suitability, MLR models were built using these wavelengths. Wavelength significance was analyzed by ANOVA, and four wavelengths (909, 951, 961, and 975 nm) were selected, setting statistical significance at the 99% confidence level. The MLR model yielded acceptable results for r² (0.92), SEP (19.97 g/l) and RMSEP (20.51 g/l). The results suggest that NIR spectroscopy is a promising technique for predicting reducing sugar content during grape ripening, as well as during the fermentation and aging of white and red wines. Individual fingerprint wavelengths strongly associated with reducing sugar content could be used to enhance the efficacy of this simple, efficient and low-cost instrument.