Search

We present the development of a bilinear regression model for multivariate calibration on the basis of maximum correntropy criteria (MCC) whose robustness can be easily controlled. MCC regression methods can be more effective when the assumption of normality does not hold or when data are contaminated with outliers. These methods are competitive when the degree of robustness against outliers should be controlled. By controlling the robustness, information from candidate outliers can be partially retained rather than completely included or discarded during calibration. Within the context of bilinear regression models, an MCC approach using statistically inspired modification of the partial least squares (SIMPLS) is proposed, which is named maximum correntropy‐weighted partial least squares (MCW‐PLS). Thanks to the controllable robustness of MCC models, observations are upweighted or downweighted during the calibration process, rendering robust models with soft discrimination of samples. Such a weighting represents an important advantage, especially for cases when samples are not drawn from a normal distribution. Applications to three real case studies are presented. These applications uncovered three main features of MCW‐PLS: robustness control between SIMPLS and robust SIMPLS (RSIMPLS), improvements in prediction performance of bilinear calibration models, and the possibility to detect the most informative samples in a calibration set. [ABSTRACT FROM AUTHOR]

In perennial ryegrass breeding programmes, dry‐matter yield (DMY) of individual plots is monitored destructively at the different cuts or derived from non‐destructive canopy height measurements using devices like rising plate meters (RPM). These approaches both have constraints. Destructive sampling implies low temporal resolution, restraining the study of dry‐matter accumulation rates, while RPM measurements are influenced by the canopy structure and limit intra‐field variability identification. We present a phenotyping methodology, based on the use of an affordable RGB camera mounted on an unmanned aerial vehicle (UAV), to monitor the spatial and temporal evolution of canopy height and to estimate DMY. Weekly flights were carried out from April to October above a field comprising a diverse set of accessions. To test the capacity of UAV imagery to estimate canopy height, 8 ground control points and 28 artificial height references were placed at different locations. Accurate flights with an RMSE as low as 0.94 cm were achieved. In addition, canopy height was recorded using an RPM and destructive biomass samples were collected. Different models (linear, multiple linear, principal components, partial least squares regression and random forest) were used to predict DMY, and their performance was evaluated. The best estimations were obtained by combining variables including canopy height, vegetation indices and environmental data in a multiple linear regression (R2 = .81). All models built using UAV data obtained a lower RMSE than the one using RPM data. The approach presented is a possibility for breeders to incorporate new information in their selection process. [ABSTRACT FROM AUTHOR]

[Display omitted] • The principles of four methods for food quality analysis are discussed. • Different MVDA methods are illustrated on a common data set as an example-based review. • The most suitable MVDA method depends on the nature of the data set and the objective of the research. • The potential of more advanced multivariate methods should be further explored. These days, large multivariate data sets are common in the food research area. This is not surprising as food quality, which is important for consumers, and its changes are the result of a complex interplay of multiple compounds and reactions. In order to comprehensively extract information from these data sets, proper data analysis tools should be applied. The application of multivariate data analysis (MVDA) is therefore highly recommended. However, at present the use of MVDA for food quality investigations is not yet fully explored. This paper focusses on a number of MVDA methods (PCA (Principal Component Analysis), PLS (Partial Least Squares Regression), PARAFAC (Parallel Factor Analysis) and ASCA (ANOVA Simultaneous Component Analysis)) useful for food quality investigations. The terminology, main steps and the theoretical basis of each method will be explained. As this is an example-based review, each method was applied on the same experimental data set to give the reader an idea about each selected MVDA method and to make a comparison between the outcomes. Numerous MVDA methods are available in literature. Which method to select depends on the data set and objective. PCA should be the first choice for data exploration of two-dimensional data. For predictive purposes, PLS is the most appropriate method. Given an underlying experimental design, ASCA takes into account both the relation between the different variables and the design factors. In case of a multi-way data set, PARAFAC can be used for data exploration. While these methods have already proven their value in research, there is a need to further explore their potential to investigate the complex interplay of compounds and reactions contributing to food quality. With this work we would like to encourage food scientists with no or limited knowledge of MVDA to get some first insights into the selected methods. [ABSTRACT FROM AUTHOR]

Spectral data from multiple sources can be integrated into multi-block fusion chemometric models, such as sequentially orthogonalized partial-least squares (SO-PLS), to improve the prediction of sample quality features. Pre-processing techniques are often applied to mitigate extraneous variability, unrelated to the response variables. However, the selection of suitable pre-processing methods and identification of informative data blocks becomes increasingly complex and time-consuming when dealing with a large number of blocks. The problem addressed in this work is the efficient pre-processing, selection, and ordering of data blocks for targeted applications in SO-PLS. We introduce the PROSAC–SO–PLS methodology, which employs pre-processing ensembles with response-oriented sequential alternation calibration (PROSAC). This approach identifies the best pre-processed data blocks and their sequential order for specific SO-PLS applications. The method uses a stepwise forward selection strategy, facilitated by the rapid Gram-Schmidt process, to prioritize blocks based on their effectiveness in minimizing prediction error, as indicated by the lowest prediction residuals. To validate the efficacy of our approach, we showcase the outcomes of three empirical near-infrared (NIR) datasets. Comparative analyses were performed against partial-least-squares (PLS) regressions on single-block pre-processed datasets and a methodology relying solely on PROSAC. The PROSAC–SO–PLS approach consistently outperformed these methods, yielding significantly lower prediction errors. This has been evidenced by a reduction in the root-mean-squared error of prediction (RMSEP) ranging from 5 to 25 % across seven out of the eight response variables analyzed. The PROSAC–SO–PLS methodology offers a versatile and efficient technique for ensemble pre-processing in NIR data modeling. It enables the use of SO-PLS minimizing concerns about pre-processing sequence or block order and effectively manages a large number of data blocks. This innovation significantly streamlines the data pre-processing and model-building processes, enhancing the accuracy and efficiency of chemometric models. Summary of the PROSAC–SO–PLS methodology applied to one of the response variables (protein) of the miniS-milk dataset, which is one of the three datasets presented in this study. In this specific case, multiple blocks of spectral data derived from multiple miniature spectrometers measuring the same milk samples simultaneously are subjected to an ensemble of pre-processing techniques within the PROSAC framework. The five most variance-explaining blocks identified by PROSAC are subsequently utilized as input variables for SO-PLS, facilitating the construction of a model that optimizes block utilization for predicting the protein content in milk, with protein content reference values obtained from laboratory analyses. [Display omitted] • PROSAC and SO-PLS are combined for improved multi-block data fusion. • PROSAC streamlines SO-PLS by easing pre-processing and block order concerns. • PROSAC–SO–PLS effectively manages a large number of input data blocks. • PROSAC–SO–PLS is tested on three NIR datasets, surpassing existing state-of-art. • PROSAC–SO–PLS reduces prediction errors by 5–25 % for the evaluated NIR datasets. [ABSTRACT FROM AUTHOR]

Banana and its dried products are one of the most common foods consumed all over the world. Quality attributes of dried banana slices are typically evaluated using time-consuming and labor intensive methods which do not allow to evaluate each individual slice. Replacement of the current destructive methods for quality evaluation of dried banana slices by fast, non-destructive methods would provide large added value to the food industry. Therefore, the aim of this study was to monitor the moisture content, texture and color of banana slices during the drying process in a fast and non-destructive way. VNIR hyperspectral reflectance imaging in the 400–1000 nm range was selected for this purpose. Thanks to a cross-polarized configuration the effects of glare or specular reflection on the banana slice surfaces in the hyperspectral diffuse reflectance images was largely reduced. Reference quality attributes for all the slices (moisture content, texture and color ( L*, a*, b* values) obtained using conventional destructive methods and colorimeter were predicted from their corresponding average reflectance spectra extracted from the hyperspectral images by means of partial least squares regression (PLSR). The PLSR calibration models were validated on samples which had not been used for model calibration. The results were very good for water content ( R 2 P  = 0.97, RMSEP  = 0.05 kg water/kg DM), quite good for and b* value ( R 2 P  = 0.83, RMSEP  = 1.95), and reasonable for texture ( R 2 P  = 0.66, RMSEP  = 11.8 N), a* value ( R 2 P  = 0.53, RMSEP  = 1.32) and L* value ( R 2 P  = 0.61, RMSEP  = 5.92). Subsequently, these calibration models were used to predict those quality attributes at pixel level for the validation slices to visualize the spatial distribution of these quality parameters at different stages during the drying process. The obtained results clearly indicate the potential of cross-polarized hyperspectral reflectance imaging for non-destructive monitoring of the quality attributes of banana slices during drying. [ABSTRACT FROM AUTHOR]

[Display omitted] • Milk composition • Calibration maintenance using unsupervised bias monitoring • Near-infrared spectroscopy • Partial least squares • Bulk milk analysis Long-term studies have shown a bias drift over time in the prediction performance of near-infrared spectroscopy measurement systems. This bias drift generally requires extra laboratory reference measurements to detect and correct for this bias. Since these reference measurements are expensive and time consuming, there is a need for advanced methodologies for bias drift monitoring and correction without the need for taking extra samples. In this study, we propose and validate a method to monitor the bias drift and two methods to tackle it. The first method requires no extra measurements and uses a modified version of Partial Least Squares Regression to estimate and correct the bias. This method is based on the assumption that the mean concentration of the predicted component remains constant over time. The second method uses regular bulk milk measurements as a reference for bias correction. This method compares the measured concentrations of the bulk milk to the volume-weighted average concentrations of individual milk samples predicted by the sensor. Any difference between the actual and calculated bulk milk composition is then used to perform a bias correction on the predictions by the sensor system. The effectiveness of these methods to improve the component prediction was evaluated on data originating from a custom-built sensor that automatically measures the NIR reflectance and transmittance spectra of raw milk on the farm. We evaluate the practical use case where models for predicting the milk composition are trained upon installation of the sensor at the farm, and later used to predict the composition of subsequent samples over a period of more than 6 months. The effectiveness of the fully unsupervised method was confirmed when the mean concentration of the milk samples remained constant, while the effectiveness reduced when this was not the case. The bulk milk correction method was effective when all relevant samples for the component were measured by the sensor and included in the analyzed bulk milk, but is less effective when samples included in the bulk which are not measured by the sensor system. When the necessary conditions are met, these methods can be used to extend the lifetime of deployed prediction models by significantly reducing the bias on the predicted values. [ABSTRACT FROM AUTHOR]

The sugar profile of a syrup determines its suitability for use in different food applications. The traditional wet chemical method for sugar profile analysis is time-consuming, destructive, and requires several chemical reagents. Therefore, this study aimed to develop a rapid method for determining the concentration of the main sugars (glucose, maltose, maltotriose, and fructose) in tapioca syrups using NIR spectroscopy in the 12000–4000 cm−1 wavenumber region or 800–2500 nm wavelength range. As large spectral variation due to water and liquid glucose was observed, orthogonal projection (OP) was evaluated to reduce those effects. Results showed that for all sugars studied, orthogonal projection prior to partial least squares regression (PLSR) improved the prediction performance in terms of the root mean square error of prediction (RMSEP) by 5–47% and the coefficient of determination for prediction (R p 2) by 3–7%. The developed models had high R p 2 and low RMSEP values, which provides opportunities for industrial monitoring of the sugar profiles of tapioca syrup products. • NIR spectroscopy allows quantifying many sugars in tapioca syrups. • Interference removal by orthogonal projection enhances prediction of syrup sugars. • High-accuracy prediction models for the individual sugars were built by using PLSR. • Identification of important wavelengths for syrup sugar profiling was established. [ABSTRACT FROM AUTHOR]

In high-yielding dairy cattle, severe postpartum negative energy balance is often associated with metabolic and infectious disorders that negatively affect production, fertility, and welfare. Mobilization of adipose tissue associated with negative energy balance is reflected through an increased level of nonesterified fatty acids (NEFA) in the blood plasma. Earlier, identification of negative energy balance through detection of increased blood plasma NEFA concentration required laborious and stressful blood sampling. More recently, attempts have been made to predict blood NEFA concentration from milk samples. In this study, we aimed to develop and validate a model to predict blood plasma NEFA concentration using the milk mid-infrared (MIR) spectra that are routinely measured in the context of milk recording. To this end, blood plasma and milk samples were collected in wk 2, 3, and 20 postpartum for 192 lactations in 3 herds. The blood plasma samples were taken in the morning, and representative milk samples were collected during the morning and evening milk sessions on the same day. To predict plasma NEFA concentration from the milk MIR spectra, partial least squares regression models were trained on part of the observations from the first herd. The models were then thoroughly validated on all other observations of the first herd and on the observations of the 2 independent herds to explore their robustness and wide applicability. The final model could accurately predict blood plasma NEFA concentrations <0.6 mmol/L with a root mean square error of prediction of <0.143 mmol/L. However, for blood plasma with >1.2 mmol/L NEFA, the model clearly underestimated the true level. Additionally, we found that morning blood plasma NEFA levels were predicted with significantly higher accuracy using MIR spectra of evening milk samples compared with MIR spectra of morning samples, with root mean square error of prediction values of, respectively, 0.182 and 0.197 mmol/L, and R2 values of 0.613 and 0.502. These results suggest a time delay between variations in blood plasma NEFA and related milk biomarkers. Based on the MIR spectra of evening milk samples, cows at risk for negative energy status, indicated by detrimental morning blood plasma NEFA levels (>0.6 mmol/L), could be identified with a sensitivity and specificity of, respectively, 0.831 and 0.800. As this model can be applied to millions of historical and future milk MIR spectra, it opens an opportunity for regular metabolic screening and improved resilience phenotyping. [ABSTRACT FROM AUTHOR]

Optical measurement of fruit quality can provide the specific insights into the light absorption (μ a) of chemical compositions and the multiple scattering (μ s ′) by structured tissues, and contributes to their precise evaluation. This work comprehensively explored the bulk optical properties (BOPs) in the wavelength range of 600–1500 nm related to the soluble sugars, pectin constitution and microstructural features (area, perimeter, Feret's diameter, cell roundness, cell solidity and intercellular space) of 'Korla' pear cortex during post-harvest storage. The high correlation between μ a values around 980 nm and 1 485 nm and total sugars and sucrose contents were observed with the r values from 0.74 to 0.99. The μ s ′ of pear tissues were positively corelated to the firmness, cell microstructure parameters in cortex (area, perimeter, Feret's diameter and solidity), whereas negatively correlated to their cell roundness and intercellular properties. Besides, the partial least squares regression models based on the featured BOPs of pear cortex successfully predicted the total sugars and sucrose contents with the determination coefficients (R p 2) from 0.73 to 0.91. Further, the combined BOPs feature models based on the first component (PC1) scores of BOPs provide excellent prediction of firmness (R p 2 = 0.90) and acid soluble pectin (R p 2 = 0.93). Accordingly, this study indicated the optical properties have the possibility to present the sugar composition, firmness, pectin and microstructural characteristics of pear tissues. • Sugars and microstructure of pear tissues are reflected in optical properties (BOPs). • The total sugars and sucrose contribute to assess soluble solids in pears by BOPs. • The acid-soluble pectin and cell solidity are key to assess pear structure by BOPs. • The first component scores on BOPs improved the prediction of pear microstructure. [ABSTRACT FROM AUTHOR]

• Bean type had a significant effect on tannins contents. • Phytate to iron and phytate to zinc molar ratios of raw beans were high. • NIR spectroscopy predicted antinutrient to mineral molar ratios of raw beans. • Ability to predict antinutrient influenced prediction of molar ratios using NIR. The presence of antinutrients in common beans negatively affects mineral bioavailability. Therefore, this study aimed to predict the antinutrient to mineral molar ratios (proxy-indicators of in vitro mineral bioavailability) of a wide range of raw bean types, using near-infrared (NIR) spectroscopy. Iron, zinc, phytate and tannin concentrations and, antinutrient to mineral molar ratios were determined. Next, model calibration using NIR spectra from milled beans was performed. This entailed wavelength selection, pre-processing and partial least squares regression. Bean type had a significant effect on tannin content. The average values of phytate to iron (Phy:Fe), phytate to zinc (Phy:Zn), tannins to iron (Tan:Fe) and phytate and tannins to iron (Phy + Tan:Fe) MRs were 27.6, 61.7, 16.0 and 43.6, respectively. With determination coefficients for test set prediction above 75%, the PLS-R models for Phy:Zn, Tan:Fe and Phy + Tan:Fe molar ratios are useful for screening purposes. [ABSTRACT FROM AUTHOR]

Indirect quantification of chemical composition through spectral measurements requires the establishment of multivariate calibration models. The reference analyses on the calibration samples typically form a major cost factor in the establishment of these multivariate models. Therefore, the aim of this study was to select the most informative calibration samples in an unsupervised way based on the spectral measurements. To this end, guidelines to address this challenge in PLSR model building have been developed. The recommendations include calculating a sample size that surpasses the model complexity by a factor of 12, performing the selection in the PCA score space spanned by a sufficiently large number of principal components and using methods such as Kennard-Stone, Puchwein, Clustering or D-optimal designs. We provide the data and methodology used in the present study for future use. • We aim to find a subset that would render a model of comparable performance to the model rendered by the full set of samples. • The optimal sample size for a multivariate calibration model depends on the complexity of such a model. • It is possible to evaluate the quality of the selected samples by comparing covariance matrices for PLSR calibration models. • Control over the sample size and the input dimensionality lowers the impact of the selection method. [ABSTRACT FROM AUTHOR]

The cooking time of common beans is influenced by genotype and storage conditions. This study aimed to use near-infrared (NIR) spectra of milled, freshly harvested (fresh) beans to predict their cooking times and their susceptibility to develop the storage-induced, hard-to-cook (HTC) defect. The physical characteristics of bean accessions, from two different seasons, were evaluated. The cooking times and susceptibility to HTC (determined by aging under standard adverse conditions) of the aforesaid beans were correlated to NIR spectra to develop calibrations using variable selection and partial least squares regression. The beans exhibited diverse physical characteristics, cooking times and susceptibility to HTC. The models predicting cooking times of fresh beans sufficiently overcame genotype and seasonal differences (R2 P = 0.73, RMSEP = 4 min). The susceptibility of fresh beans to HTC was also successfully predicted (R2 CV = 0.8, RSECV = 64%). NIR spectroscopy has high potential to rapidly identify beans with short cooking times and low susceptibility to HTC at harvest. • Bean type had a significant effect on the physical characteristics of the accessions. • Cooking times of fresh beans were predicted from NIR spectra of their flours. • PLS-R models were robust against phenotypic and seasonal variations of the beans. • Susceptibility to develop HTC was predicted from NIR spectra of milled, fresh beans. [ABSTRACT FROM AUTHOR]

The cooking time of beans is an important quality indicator which can change considerably during ageing. Therefore, this study investigated the potential of near-infrared spectroscopy to rapidly predict cooking times of aged common beans. Four bean varieties were aged under different storage conditions, resulting in a range of samples for each of these varieties. The cooking kinetics of the aged beans were determined by finger pressing and modelled using logistic regression to obtain the times it took to cook 95% of the beans. The cooking times obtained were predicted from the NIR spectra of milled raw bean samples. This was done using partial least squares regression, after carrying out wavelength selection. Model performance was improved up to an average prediction error of 8 min by de-hulling the beans and reducing the number of varieties included. In conclusion, NIR spectroscopy has high potential to predict the cooking times of aged beans. • Cooking times of beans were predicted from NIR spectra of milled raw beans. • De-hulling beans prior to NIR measurements improved prediction accuracy. • NIR spectra of aged beans clustered into groups according to variety. • Variety-specific calibrations more accurately predict cooking times of aged beans. [ABSTRACT FROM AUTHOR]

• Contrast enhanced micro-CT was refined for surface microstructure data of apple. • Reduced scattering coefficient extracted by Hyperspectral Laser Scattering Imagin.g • PLS model built to determine the microstructural properties affecting scattering. • Overall porosity and pore surface density related well with scattering (R2 = 0.89). • Optical determination of subsurface porosity seems feasible. The success of long-term storage of apples under controlled atmosphere (CA) depends, amongst others, on the gas exchange properties of the fruit. As gas exchange is effectively dictated by the microstructure of the fruit, the ability to obtain microstructure data becomes critical to improve storage solutions. The current study complements scattering measurements by means of spatially resolved spectroscopy (SRS) with 3D microstructure data obtained with contrast enhanced X-ray computed micro-tomography (micro-CT). Complementary measurements with both techniques were performed on apples of different cultivars that have different optimal storage conditions and different fleshy microstructure ('Kanzi', 'Braeburn', 'Jonagold' and 'Golden Delicious'). The mean reduced scattering coefficients of the subsurface tissue were calculated from SRS measurements in the 750 nm to 900 nm range at specific equatorial positions on the intact apple. Microstructural parameters such as cell size (equivalent spherical diameter), anisotropy, elongation, flatness, sphericity, object count, porosity as well as pore surface density were quantified and analyzed at the same spot up to 3 mm in depth from the fruit surface. A partial least squares regression model using the microstructural parameters as the different variables to predict the reduced scattering coefficient was built in order to identify the parameters contributing most to this relation. Both mean porosity and pore surface density showed the largest absolute regression coefficients. Furthermore, plotting the reduced scattering coefficient against mean porosity and pore surface density produced a linear relationship between the two parameters with an R2 of 0.89 for both sets of data. The linear relationship suggests that the porosity of individual fruit can be determined via optical SRS measurements. This could allow to sort fruit based on their porosity, thus promoting fine tuning of the storage strategies by reducing variation in the porosity and gas exchange rate of the fruit being stored together. [ABSTRACT FROM AUTHOR]