Use of artificial neural networks in near-infrared reflectance spectroscopy calibrations for predicting the inclusion percentages of wheat and sunflower meal in compound feedingstuffs

Deep Learning-Based Quantitative Assessment of Melamine and Cyanuric Acid in Pet Food Using Fourier Transform Infrared Spectroscopy

Melamine and its derivative, cyanuric acid, are occasionally added to pet meals because of their nitrogen-rich qualities, leading to the development of several health-related issues. A nondestructive sensing technique that offers effective detection must be developed to address this problem. In conjunction with machine learning and deep learning technique, Fourier transform infrared (FT-IR) spectroscopy was employed in this investigation for the nondestructive quantitative measurement of eight different concentrations of melamine and cyanuric acid added to pet food. The effectiveness of the one-dimensional convolutional neural network (1D CNN) technique was compared with that of partial least squares regression (PLSR), principal component regression (PCR), and a net analyte signal (NAS)-based methodology, called hybrid linear analysis (HLA/GO). The 1D CNN model developed for the FT-IR spectra attained correlation coefficients of 0.995 and 0.994 and root mean square error of prediction values of 0.090% and 0.110% for the prediction datasets on the melamine- and cyanuric acid-contaminated pet food samples, respectively, which were superior to those of the PLSR and PCR models. Therefore, when FT-IR spectroscopy is employed in conjunction with a 1D CNN model, it serves as a potentially rapid and nondestructive method for identifying toxic chemicals added to pet food.

The Application of NIRS to Determine Animal Physiological Traits for Wildlife Management and Conservation

The ability to measure and monitor wildlife populations is important for species management and conservation. The use of near-infrared spectroscopy (NIRS) to rapidly detect physiological traits from wildlife scat and other body materials could play an important role in the conservation of species. Previous research has demonstrated the potential for NIRS to detect diseases such as the novel COVID-19 from saliva, parasites from feces, and numerous other traits from animal skin, hair, and scat, such as cortisol metabolites, diet quality, sex, and reproductive status, that may be useful for population monitoring. Models developed from NIRS data use light reflected from a sample to relate the variation in the sample’s spectra to variation in a trait, which can then be used to predict that trait in unknown samples based on their spectra. The modelling process involves calibration, validation, and evaluation. Data sampling, pre-treatments, and the selection of training and testing datasets can impact model performance. We review the use of NIRS for measuring physiological traits in animals that may be useful for wildlife management and conservation and suggest future research to advance the application of NIRS for this purpose.

LOCAL regression applied to a citrus multispecies library to assess chemical quality parameters using near infrared spectroscopy

The non-destructive on-tree measurement of the chemical quality attributes of fruits belonging to the Citrus genus using rapid spectral sensors is of vital interest to citrus growers, allowing them to carry out a selective harvest of any species of Citrus fruit. With this objective, the viability of using of a handheld portable near infrared spectroscopy (NIRS) instrument to predict soluble solid content (SSC), pH, titratable acidity (TA), maturity index and BrimA, in order to measure the optimum harvest time in a group made up of 608 samples belonging to the Citrus genus (378 oranges and 230 mandarins) was evaluated. For each of the parameters analysed, both non-linear regression (LOCAL algorithm) and linear regression (Modified Partial Least Squares, MPLS) strategies were designed and compared. The use of the LOCAL algorithm in the sample group of oranges and mandarins for all the parameters analysed allowed to obtain more robust models than those obtained with MPLS regression, and it could also be extended more easily when routinely applied. The results confirm that NIRS technology combined with non-linear regression strategies such as the LOCAL algorithm can indeed respond to the needs of the Citrus growers and help them to set the optimum harvest time, in this case of oranges and mandarins, by predicting the chemical quality parameters in situ.

Combination of optical coherence tomography and near infrared spectroscopy enhances determination of articular cartilage composition and structure

Conventional arthroscopic evaluation of articular cartilage is subjective and poorly reproducible. Therefore, implementation of quantitative diagnostic techniques, such as near infrared spectroscopy (NIRS) and optical coherence tomography (OCT), is essential. Locations (n = 44) with various cartilage conditions were selected from mature equine fetlock joints (n = 5). These locations and their surroundings were measured with NIRS and OCT (n = 530). As a reference, cartilage proteoglycan (PG) and collagen contents, and collagen network organization were determined using quantitative microscopy. Additionally, lesion severity visualized in OCT images was graded with an automatic algorithm according to International Cartilage Research Society (ICRS) scoring system. Artificial neural network with variable selection was then employed to predict cartilage composition in the superficial and deep zones from NIRS data, and the performance of two models, generalized (including all samples) and condition-specific models (based on ICRS-grades), was compared. Spectral data correlated significantly (p < 0.002) with PG and collagen contents, and collagen orientation in the superficial and deep zones. The combination of NIRS and OCT provided the most reliable outcome, with condition-specific models having lower prediction errors (9.2%) compared to generalized models (10.4%). Therefore, the results highlight the potential of combining both modalities for comprehensive evaluation of cartilage during arthroscopy.

Evaluation of local approaches to obtain accurate near-infrared (NIR) equations for prediction of ingredient composition of compound feeds

This research work investigated new methods to improve the accuracy of intact feed calibrations for the near-infrared (NIR) prediction of the ingredient composition. When NIR reflection spectroscopy, together with linear models, was used for the prediction of the ingredient composition, the results were not always acceptable. Therefore, other methods have been investigated. Three different local methods (comparison analysis using restructured near-infrared and constituent data [CARNAC]), locally weighed regression [LWR], and LOCAL) were applied to a large (N = 20 320) and heterogeneous population of non-milled feed compounds for the NIR prediction of the inclusion percentage of wheat and sunflower meal, as representative of two different classes of ingredients. Compared with partial least-squares regression, results showed considerable reductions of standard error of prediction values for all methods and ingredients: reductions of 59, 47, and 50% with CARNAC, LWR, and LOCAL, respectively, for wheat, and reductions of 49, 45, and 43% with CARNAC, LWR, and LOCAL, respectively, for sunflower meal. These results are a valuable achievement in coping with legislation and manufacture requirements concerning the labeling of intact feedstuffs.

Predicting the age and type of tuocha tea by fourier transform infrared spectroscopy and chemometric data analysis

Fourier transform infrared (FTIR) spectroscopy combined with chemometric multivariate methods was proposed to discriminate the type (unfermented and fermented) and predict the age of tuocha tea. Transmittance FTIR spectra ranging from 400 to 4000 cm(-1) of 80 fermented and 98 unfermented tea samples from Yunnan province of China were measured. Sample preparation involved finely grinding tea samples and formation of thin KBr disks (under 120 kg/cm(2) for 5 min). For data analysis, partial least-squares (PLS) discriminant analysis (PLSDA) was applied to discriminate unfermented and fermented teas. The sensitivity and specificity of PLSDA with first-derivative spectra were 93 and 96%, respectively. Multivariate calibration models were developed to predict the age of fermented and unfermented teas. Different options of data preprocessing and calibration models were investigated. Whereas linear PLS based on standard normal variate (SNV) spectra was adequate for modeling the age of unfermented tea samples (RMSEP = 1.47 months), a nonlinear back-propagation-artificial neutral network was required for calibrating the age of fermented tea (RMSEP = 1.67 months with second-derivative spectra). For type discrimination and calibration of tea age, SNV and derivative preprocessing played an important role in reducing the spectral variations caused by scattering effects and baseline shifts.

Multivariate near-infrared reflection spectroscopy strategies for ensuring correct labeling at feed bagging in the animal feed industry

A key concern in animal feed factories is guaranteeing the correct labeling of compound feeds. Therefore, due to incorrect labeling, there is an urgent need for new control methods on the claims that can be made. In this study, this question has been tackled with different multivariate classification algorithms based on the near-infrared spectral fingerprint obtained from a given compound feed analyzed in its original physical market presentation form (i.e., cubes, coarse meals, pellets). The objective of this paper is the evaluation of different methods for establishing a separation among 24 feed types. Two linear methods, soft independent modeling of class analogy (SIMCA) and partial least squares (PLS) with two approaches to classification (PLSD and PLS-LDA); and one nonlinear method, support vector machines (SVM), were studied. The database used had the following structure: a first division was made between granules and meals; within these two groups, there was a second division according to three animal species to which the feed was marketed (bovine, ovine, and porcine); within each species there was a third division according to the age or physiological status of the animal (i.e., lactating dairy cattle, starters, etc.). Given the database structure, all the methods were evaluated following two strategies: (1) development of a model composed of the nine classification models corresponding to the structure of the data; and (2) development of a unique model that discriminates among the 24 classes of different feeds. With both strategies the lowest percentage of misclassified samples was achieved with the SVM method (3.96% with strategy 1 and 2.31% with strategy 2). Among the linear methods evaluated, SIMCA yielded the best results, with a percentage of 8.47% misclassified samples with strategy 1 and 4.05% misclassified samples with strategy 2. The results in this study show the ability of near-infrared spectroscopy to make acceptable classifications of feed types based only on spectral information, with differences in performance depending on the multivariate algorithm used.

Advanced nonlinear approaches for predicting the ingredient composition in compound feedingstuffs by near-infrared reflection spectroscopy

For quantitative applications, the most common usage of near-infrared reflection spectroscopy (NIRS) technology, calibration involves establishing a mathematical relationship between spectral data and data provided by the reference. This model may be fairly complex, since the near-infrared spectrum is highly variable and contains physical/chemical information for the sample that may be redundant, and multivariate calibration is usually required. When the relationship to be modeled is nonlinear, classical regression methods are inadequate, and more complex strategies and algorithms must be sought in order to model this nonlinearity. The development of NIRS calibrations to predict the ingredient composition, i.e., the inclusion percentage of each ingredient, in compound feeds is a complex task, due to the nature of the parameters to be predicted and to the heterogeneous nature of the matrices/formulas in which each ingredient participates. The present paper evaluates the use of least squares support vector machines (LSSVM) and two local calibration methods, CARNAC and locally biased regression, for developing NIRS models to predict two of the most representative ingredients in compound feed formulations, wheat and sunflower meal, using a large spectral library of 7523 commercial compound feed samples. For both ingredients, the best results were obtained using CARNAC, with standard errors of prediction (SEP) of 1.7% and 0.60% for wheat and sunflower meal, respectively, and even better results when the algorithm was allowed to refuse to predict 10% of the unknowns. Meanwhile, LSSVM performed less well on wheat (SEP 2.6%) but comparably on sunflower meal (SEP 0.60%), giving results very similar to those reported previously for artificial neural networks. Locally biased regression was the least successful of the three methods, with SEPs of 3.3% for wheat and 0.72% for sunflower meal. All the nonlinear methods improved on the standard approach using partial least squares (PLS), which gave SEPs of 5.3% for wheat and 0.81% for sunflower meal.