FT-NIR Data Research Articles

Development of automatic tuning for combined preprocessing and hyperparameters of machine learning and its application to NIR spectral data of coconut milk adulteration

This study proposed a novel approach to automatically select the preprocessing methods and hyperparameters of machine learning (ML) algorithms based on their best performance in cross-validation for near-infrared (NIR) spectroscopy data. The proposed method simultaneously incorporates single or multiple-preprocessing steps and tunes hyperparameters to determine the best model performance for FT-NIR and Micro-NIR spectral data of coconut milk adulteration with distilled water and mature coconut water in the range of 0%–50%. Computational experiments were conducted using nine single preprocessing types, three types of ML classifier (linear discriminant analysis (LDA), k-nearest neighbour (KNN), multilayer perceptron (MLP)) and three types of ML regressor (partial least squares (PLS), KNN, MLP). The proposed performance strategy effectively addressed and produced satisfactory outcomes for classification and regression challenges in coconut milk adulteration. Finally, the results demonstrated that the proposed approach can more accurately determine the best model, particularly for NIR spectroscopy of coconut milk adulteration.

Gasoline discrimination using infrared spectroscopy and virtual samples based on measurement uncertainty

In a previous work, we proposed a methodology for pair-wise discrimination of gasoline samples by creating virtual samples based on physicochemical assays or distillation curves. Satisfactory results were achieved, although specialist and specific apparatus (not commonly available at police laboratories) were required. The present study goes a step further and for the first time investigates the possibility of infrared (IR) spectroscopy to enable a virtual samples-based methodology for comparison of gasoline samples in pairs. IR spectroscopy feasibility for in situ applications is attractive for forensic investigations. The performances of one handheld NIR device and one dual-range (FT-NIR and FT-IR) benchtop spectrometer were evaluated. The estimation of uncertainty in infrared spectral measurement (needed to generate virtual samples) is barely discussed in literature. So far, there are no literature reports describing quantification and comparison of measurement uncertainties for the spectral acquisitions evaluated here, especially regarding their use for generating virtual samples. A stepwise procedure to quantify uncertainties associated with IR spectral acquisition, at each wavenumber, is described. This method can be useful for understanding both the sources of variability in IR measurements and the system under investigation. Uncertainty estimation was based on experimental data and considered intermediate precision, repeatability and variations in sample temperature as sources of variability. Virtual samples were employed in a discrimination approach using SIMCA models. Results for portable NIR, FT-NIR and FT-IR data sets showed complete discrimination for 96.3%, 93.4% and 93.7% of the 1431 pairs of gasoline samples evaluated, respectively. These results were comparable and similar to those obtained for the physicochemical properties data set (95.7%), although slightly inferior to the result obtained for distillation curves (99.2%). Using IR non-destructive methods in this case could enable faster investigations and simpler analysis, especially for the low-cost handheld spectrometer. In a screening approach, atmospheric distillation assays can be employed only if infrared techniques are not capable of distinguishing the samples subject to comparison. In this work, a pair of samples was considered to be completely discriminated only when a null false positive error (FPR) was achieved, although a more flexible criterium may be acceptable in practice. Finally, the methodology could be extended to other applications where sample comparison is important.

Rapid Discrimination of the Country Origin of Soybeans Based on FT-NIR Spectroscopy and Data Expansion

Rapid Discrimination of the Country Origin of Soybeans Based on FT-NIR Spectroscopy and Data Expansion

In Situ Monitoring of Sugar Content in Breakfast Cereals Using a Novel FT-NIR Spectrometer

This research demonstrates simultaneous predictions of individual and total sugars in breakfast cereals using a novel, handheld near-infrared (NIR) spectroscopic sensor. This miniaturized, battery-operated unit based on Fourier Transform (FT)-NIR was used to collect spectra from both ground and intact breakfast cereal samples, followed by real-time wireless data transfer to a commercial tablet for chemometric processing. A total of 164 breakfast cereal samples (60 store-bought and 104 provided by a snack food company) were tested. Reference analysis for the individual (sucrose, glucose, and fructose) and total sugar contents used high-performance liquid chromatography (HPLC). Chemometric prediction models were generated using partial least square regression (PLSR) by combining the HPLC reference analysis data and FT-NIR spectra, and associated calibration models were externally validated through an independent data set. These multivariate models showed excellent correlation (Rpre ≥ 0.93) and low standard error of prediction (SEP ≤ 2.4 g/100 g) between the predicted and the measured sugar values. Analysis results from the FT-NIR data, confirmed by the reference techniques, showed that eight store-bought cereal samples out of 60 (13%) were not compliant with the total sugar content declaration. The results suggest that the FT-NIR prototype can provide reliable analysis for the snack food manufacturers for on-site analysis.

Assessing the interaction between drying and addition of maltodextrin to Kakadu plum powder samples by two dimensional and near infrared spectroscopy.

The effect of drying (oven and freeze-drying) and the addition of maltodextrin as a carrier to Kakadu plum (Terminalia ferdinandiana) puree powder samples were evaluated using a combination of two dimensional (2DCOS) and near infrared (NIR) spectroscopy. Fruit powder samples were obtained from an experiment where oven and freeze-drying methods were compared together with the addition of seven levels of maltodextrin to the samples (control, 5, 7.5, 10, 15, 20 and 25% w/w). Samples were scanned using a FT-NIR instrument (Tango, Bruker, Germany) and data analysed using 2DCOS. Asynchronous and synchronous 2DCOS spectroscopy were used to analyse and interpret the effects of the method of drying and the addition of maltodextrin on the NIR spectra of the fruit samples. The utilization of 2DCOS combined with NIR spectroscopy showed how the drying method affect the NIR spectra and thus the main implications of developing an effective, quick, and easy to use protocol for determining the drying method.

FT-NIR and linear discriminant analysis to classify chickpea seeds produced with harvest aid chemicals

Spectroscopy and machine learning (ML) algorithms have provided significant advances to the modern food industry. Instruments focusing on near-infrared spectroscopy allow obtaining information about seed and grain chemical composition, which can be related to changes caused by field pesticides. We investigated the potential of FT-NIR spectroscopy combined with Linear Discriminant Analysis (LDA) to discriminate chickpea seeds produced using different desiccant herbicides at harvest anticipation. Five herbicides applied at three moments of the plant reproductive stage were utilized. The NIR spectra obtained from individual seeds were used to build ML models based on LDA algorithm. The models developed to identify the herbicide and the plant phenological stage at which it was applied reached 94% in the independent validation set. Thus, the LDA models developed using near-infrared spectral data provided to be efficient, quick, non-destructive, and accurate to identify differences between seeds due to pre-harvest herbicides application.

Application of FT-NIR spectroscopy and NIR hyperspectral imaging to predict nitrogen and organic carbon contents in mine soils

Abstract The aim of this study was to compare the performance of FT-NIR spectroscopy and near-infrared hyperspectral imaging (NIR-HSI) in predicting the Corg and Nt contents in mine soils. The mine soil samples were measured for the Corg and Nt contents and their NIR spectra were recorded (1000–2500 nm). Predictive models were developed using 126 samples with partial least square regression (PLSR) or artificial neural networks (ANN) and validated with 58 independent samples. The NIR-HSI based models had distinctly higher accuracy of Corg content prediction than those based on FT-NIR data in both PLSR and ANN methods, as indicated by lower of standard errors of prediction. The prediction accuracy for the Nt content was similar for the two spectral methods and both chemometric approaches tested. The study showed that despite lower spectral resolution the NIR-HSI spectra retained all the information needed for accurate prediction of Corg and Nt contents.

Determination of Moisture in Olive Oil: Rapid FT-NIR Spectroscopic Procedure Based on the Karl-Fischer Reference Method.

According to CODEX, moisture and volatile matter are olive oil quality parameters and the development of a rapid screening method for the determination of moisture is of interest. We recently demonstrated for the first time that the weak near-infrared (NIR) band near 5260 cm-1 is primarily attributed to a water O-H combination band. To determine the intensity of this band, we measured the peak-to-peak (p-p) height of its first derivative and generated exponential calibration curves for p-p height versus gravimetrically determined concentrations of spiked water in olive oils that had been purged of their initial moisture contents. To further optimize this univariate calibration method, calibration curves were generated in the present study based on plotting the moisture band first derivative p-p heights for neat olive oils (that were neither purged nor spiked) versus the moisture concentrations obtained by the Karl-Fischer (KF) primary reference method. To enhance the speed of FT-NIR data collection, measurements were carried in the transmission mode using disposable glass tubes. We also developed and compared a multivariate partial least squares approach to the univariate one. All the spectra were collected in two separate laboratories using two FT-NIR spectrometers of the same brand and model and no significant difference (p > 0.05) was found between the two laboratory determinations and the KF reference values at a 95% confidence interval. High accuracies were found with the two FT-NIR instruments used, as indicated by the low root mean squared error (RMSE, %) for predicted values obtained with the univariate procedure (RMSE = 0.008% and 0.010%) and the multivariate one, which yielded an even lower value (RMSE= 0.007% for both instruments). These results suggest that, once validated, the FT-NIR approach could potentially be a rapid substitute for the KF method.

A modified mid-level data fusion approach on electronic nose and FT-NIR data for evaluating the effect of different storage conditions on rice germ shelf life

Evaluating the possibility of extending shelf life of rice germ (a by-product of rice milling process) by reducing water activity in combination with storage atmosphere packaging, without any heat treatment, is the aim of the present study. Samples at different water activities (0.55, 0.45 and 0.36) were packed in air, argon or under vacuum, and stored at 27 °C for 150 days. To the aim, a non-targeted approach was applied by means of an FT-NIR spectrometer in reflectance with a rotating sample holder and a portable electronic nose, equipped with 10 non-specific sensors. For understanding the impact of the factors under study on the rice germ shelf life, a modified mid-level data fusion approach was applied to enhance the information most correlated with time. Moreover, Principal Component Analysis was applied on fused data to follow samples evolution during storage and identify different clusters according to the storage conditions. The rice germ case study allowed to better understand the information captured by the non-specific sensors: a 2D correlation map was developed combining the e-nose data with the NIR spectral information, highlighting relationships among NIR absorption bands and classes of chemical compounds inducing e-nose responses. A data fusion approach highlighted the importance of water activity on rice germ storage, while no interesting differences were ascribable to storage atmosphere packaging systems. In terms of correlation, the sensors could be divided in two groups, negatively inter-correlated: sensors ascribable to aromatic compounds (WC) and correlated with the NIR band around 4800–4900 cm−1 (N–H bending of primary amides, typical for peptides coming from protein hydrolysis); broad-range response sensors (WS), linked with the NIR band at 5128 cm−1 (second overtone of CO stretching of esters).

Milk Renneting: Study of Process Factor Influences by FT-NIR Spectroscopy and Chemometrics

The dairy industry is continuously developing new strategies to obtain healthier dairy products preserving expected properties. However, when modifying a food process, the reassessment of each parameters and their interaction should be considered as highly influencing the final quality. Among others, rennet process features are fundamental for both sensory properties and typical characteristics of a cheese. In this contest, the research addresses the development of a FT-NIR spectroscopic method, coupled with chemometrics, for the study of the effect of process variables on milk renneting. The effects of temperature (30 °C, 35 °C, 40 °C), milk fat concentration (0.1, 2.55, 5 g/100 mL), and pH (6.3, 6.5, 6.7) were investigated by means of a Box-Behnken experimental design. FT-NIR data collected along the 17 trials were explored by interval-PCA (i-PCA) and ANOVA-simultaneous component analysis (ASCA). i-PCA revealed differences in the occurrence and trends of coagulation phases, related to the three considered factors. ASCA allowed the characterization of renneting evolution and the assessment of the factor role, demonstrating that main and interaction effects are significant for the process progress. The proposed approach demonstrated that i-PCA and ASCA on FT-NIR data, highlighting the effects of the operating factors, allow a rapid and accurate analysis of process modifications in cheese manufacturing.

Bee breads from two regions of Eastern Ukraine: composition, physical properties and biological activities

Five samples of polyfloral bee bread originated from two regions of Eastern Ukraine were characterised by microscopic (SEM) and spectroscopic (FT MIR ATR, FT NIR, FT Raman and Vis) methods. Identification of pollens was based on the SEM images. Spectral differences were interpreted in terms of specific contribution of the main chemical constituents (proteins, phenolics, sugars, etc.) and pigments (flavonoids and carotenoids) of bee breads. PCA of FT NIR and Vis data was used for discrimination of bee bread samples according to botanical sources. The morphometric parameters, antioxidant capacity and prebiotic ef fects of bee breads were also studied. Obtained results expand the knowledge in the composition and properties of this beekeeping product.

Tracing the Geographical Origin of Lentils (Lens culinaris Medik.) by Infrared Spectroscopy and Chemometrics

The feasibility of applying the infrared spectroscopy for the geographical origin traceability of lentils from two different countries (Italy and Canada) was investigated. In particular, lentil samples were analyzed by Fourier transform near- and mid-infrared (FT-NIR and FT-MIR) spectroscopy and then discriminated by applying supervised models, i.e., linear discriminant analysis (LDA) and partial least squares discriminant analysis (PLS-DA). To avoid LDA overfitting, two variable strategies were adopted, i.e., a variable reduction by principal component analysis and a variable compression by wavelet packet transform algorithm. FT-MIR models were more discriminating compared to FT-NIR ones with prediction abilities ranging from 98 to 100% and from 91 to 100% for cross- and external validation, respectively. The combination of the FT-MIR and FT-NIR data did not improve the model performances. These findings demonstrated the suitability of the FT-MIR spectroscopy, in combination with supervised pattern recognition techniques, to successfully classify lentils according to their geographical origin.

Near-infrared spectroscopy and X-ray fluorescence data fusion for olive leaf analysis and crop nutritional status determination

Leaf analysis is a useful way of diagnosing the nutritional status of the plants and therefore fast methods of analysis are demanded to aid in fertilization management decisions. In this work, a strategy based on the combined use of near-infrared spectroscopy (NIR) and portable energy dispersive X-Ray Fluorescence (EDXRF) is proposed as a suitable cheap and rapid alternative to traditional wet analytical methodologies. The approach has the major benefit of minimal sample preparation since leaves need to be only dried and ground. The ability of both techniques individually and applying two strategies of data fusion for the prediction of the most important plant nutrients, namely N, P, K, Ca, Mg, Mn, Zn, and B was tested. Predictive models were constructed using Partial Least Squares (PLS) to correlate the spectra with the nutrient contents. Models of unequal prediction performance in terms of the ratio of predictive deviation (RPD) were obtained for the different parameters when considering both techniques separately. Low-level data fusion, which consists of a concatenation of the raw data from both techniques, showed little improvement and even decreased the predictive ability for some elements. Better results were obtained with mid-level data fusion, that is, merging data after a feature extraction step performed by means of Principal components analysis (PCA). The results show that a fair quantitative prediction is possible for Ca, K and Mn with RPDs ≥ 2 for external validation, whereas models for N and P allowed a semiquantitative estimation. Mg and B models were less satisfactory and can be used only for distinguish between low and high levels, while Zn content cannot be predicted. Finally, the potential of the fusion of FT-NIR and EDXRF spectroscopic data for the fast screening of olive crop nutritional status has been tested. Deficiencies in important elements like N and K has been successfully detected.

Comparing the analytical performances of Micro-NIR and FT-NIR spectrometers in the evaluation of acerola fruit quality, using PLS and SVM regression algorithms

The main goal of this study was to investigate the analytical performances of a state-of-the-art device, one of the smallest dispersion NIR spectrometers on the market (MicroNIR 1700), making a critical comparison with a benchtop FT-NIR spectrometer in the evaluation of the prediction accuracy. In particular, the aim of this study was to estimate in a non-destructive manner, titratable acidity and ascorbic acid content in acerola fruit during ripening, in a view of direct applicability in field of this new miniaturised handheld device. Acerola (Malpighia emarginata DC.) is a super-fruit characterised by a considerable amount of ascorbic acid, ranging from 1.0% to 4.5%. However, during ripening, acerola colour changes and the fruit may lose as much as half of its ascorbic acid content. Because the variability of chemical parameters followed a non-strictly linear profile, two different regression algorithms were compared: PLS and SVM. Regression models obtained with Micro-NIR spectra give better results using SVM algorithm, for both ascorbic acid and titratable acidity estimation. FT-NIR data give comparable results using both SVM and PLS algorithms, with lower errors for SVM regression. The prediction ability of the two instruments was statistically compared using the Passing-Bablok regression algorithm; the outcomes are critically discussed together with the regression models, showing the suitability of the portable Micro-NIR for in field monitoring of chemical parameters of interest in acerola fruits.

Has your ancient stamp been regummed with synthetic glue? A FT-NIR and FT-Raman study

The potential of FT-NIR and FT-Raman spectroscopies to characterise the gum applied on the backside of ancient stamps was investigated for the first time. This represents a very critical issue for the collectors' market, since gum conditions heavily influence stamp quotations, and fraudulent application of synthetic gum onto damaged stamp backsides to increase their desirability is a well-documented practice. Spectral data were processed by exploratory pattern recognition tools. In particular, application of principal component analysis (PCA) revealed that both of the spectroscopic techniques provide information useful to characterise stamp gum. Examination of PCA loadings and their chemical interpretation confirmed the robustness of the outcomes. Fusion of FT-NIR and FT-Raman spectral data was performed, following both a low-level and a mid-level procedure. The results were critically compared with those obtained separately for the two spectroscopic techniques.

An experimental study on the “sequential order” rules in generalized two-dimensional correlation spectroscopy

Following our theoretical analysis on the “sequential order” rules in generalized two-dimensional (2D) correlation spectroscopy (H. Huang, Anal. Chem. 79 (2007) 8281–8292), an experimental study was conducted to test the “sequential order” rules using the FT-NIR data of poly(3-hydroxybutyrate) (PHB)/poly( l-lactic acid) (PLA) blends under uniaxial elongation and parallel polarization. The local sequential order concept proposed for the generalized two-dimensional (2D) correlation spectroscopy is now more clearly stated; “the intensity change at ν1 occurs predominantly before ν2” means that the starting time of the intensity change at ν1 is prior to that at ν2. It is this local sequential order which reflects the real and intuitive sequential order between two events in generalized situations. It has been found that the integrated/ overall sequential order results obtained from the 2D correlation analysis may be contradictory to the intuitive local sequential order. In addition, different integrated/ overall sequential orders could be obtained by selection of different sampling intervals from a certain set of experimental data, or choosing different number of the contours for the same sampling interval. These new experimental findings are a perfect reinforcement to our previous theoretical study and have further demonstrated the uncertainty of applying the “sequential order” rules in generalized 2D correlation spectroscopy.