Visible and near-infrared spectroscopic analysis of raw milk for cow health monitoring: reflectance or transmittance?

The composition of produced milk has great value for the dairy farmer. It determines the economic value of the milk and provides valuable information about the metabolism of the corresponding cow. Therefore, online measurement of milk components during milking 2 or more times per day would provide knowledge about the current health and nutritional status of each cow individually. This information provides a solid basis for optimizing cow management. The potential of visible and near-infrared (Vis/NIR) spectroscopy for predicting the fat, crude protein, lactose, and urea content of raw milk online during milking was, therefore, investigated in this study. Two measurement modes (reflectance and transmittance) and different wavelength ranges for Vis/NIR spectroscopy were evaluated and their ability to measure the milk composition online was compared. The Vis/NIR reflectance measurements allowed for very accurate monitoring of the fat and crude protein content in raw milk (R² > 0.95), but resulted in poor lactose predictions (R² < 0.75). In contrast, Vis/NIR transmittance spectra of the milk samples gave accurate fat and crude protein predictions (R² > 0.90) and useful lactose predictions (R² = 0.88). Neither Vis/NIR reflectance nor transmittance spectroscopy lead to an acceptable prediction of the milk urea content. Transmittance spectroscopy can thus be used to predict the 3 major milk components, but with lower accuracy for fat and crude protein than the reflectance mode. Moreover, the small sample thickness (1 mm) required for NIR transmittance measurement considerably complicates its online use.     

基于近红外光谱技术的鲢鱼营养成分的快速分析

目的通过采集鲢鱼的近红外光谱数据和测定鱼肉营养成分含量探索鲢鱼营养成分的快速分析方法。方法采集254个鲢鱼鱼肉样品的近红外光谱数据,经过多元散射校正、正交信号校正、数据标准化等20种方法预处理,在1000~1799 nm光谱范围内,结合化学实测值分别采用偏最小二乘法、主成分分析和BP人工神经网络技术、偏最小二乘法和BP人工神经网络技术建立鲢鱼营养成分近红外定量模型。结果鲢鱼鱼肉粗蛋白含量为12.05%~19.05%,粗脂肪含量为0.24%~5.27%,水分含量为72.62%~80.58%,灰分含量为0.46%~1.50%,数据范围较大,可满足建模要求。在3种建模方法中,近红外光谱数据结合偏最小二乘法建立的鲢鱼营养成分模型最优,所得的粗蛋白、粗脂肪、水分和灰分的近红外定量模型的相关系数分别为0.9969、0.9925、0.9831和0.9976。结论采用近红外光谱数据和偏最小二乘法建立的模型具有较好的预测能力,能较为准确、快速地分析出鲢鱼鱼肉粗蛋白、粗脂肪、水分和灰分的含量。     

基于近红外光谱技术与BP-ANN算法的豆粕品质快速检测

应用近红外漫反射光谱技术结合误差反向传递人工神经网络(BP-ANN)算法,建立豆粕品质(包括水分、粗蛋白、残油)的定量分析模型。将豆粕漫反射吸收光谱数据进行SNV、DT、SG求导、SG平滑和均值中心化处理,然后采用偏最小二乘方法(PLS)降维获取主成分,并优化选择合适的隐含层节点数、隐含层和输出层转化函数,建立校正模型,并用验证样品对校正模型进行验证。结果显示,BP-ANN法建立的水分、粗蛋白和残油的预测相关系数(R)分别为0.981、0.988、0.982,预测标准偏差(SEP)分别为0.120、0.216、0.036,均优于PLS建模方法结果,且满足传统分析方法的重复性要求,表明BP-ANN方法可用于生产过程豆粕品质的快速监控。     

Mining data from milk infrared spectroscopy to improve feed intake predictions in lactating dairy cows

Feed intake is one of the most important components of feed efficiency in dairy systems. However, it is a difficult trait to measure in commercial operations for individual cows. Milk spectrum from mid-infrared spectroscopy has been previously used to predict milk traits, and could be an alternative to predict dry matter intake (DMI). The objectives of this study were (1) to evaluate if milk spectra can improve DMI predictions based only on cow variables; (2) to compare artificial neural network (ANN) and partial least squares (PLS) predictions; and (3) to evaluate if wavelength (WL) selection through Bayesian network (BN) improves prediction quality. Milk samples (n = 1,279) from 308 mid-lactation dairy cows [127 ± 27 d in milk (DIM)] were collected between 2014 and 2016. For each milk spectra time point, DMI (kg/d), body weight (BW, kg), milk yield (MY, kg/d), fat (%), protein (%), lactose (%), and actual DIM were recorded. The DMI was predicted with ANN and PLS using different combinations of explanatory variables. Such combinations, called covariate sets, were as follows: set 1 (MY, BW^0.75, DIM, and 361 WL); set 2 [MY, BW^0.75, DIM, and 33 WL (WL selected by BN)]; set 3 (MY, BW^0.75, DIM, and fat, protein, and lactose concentrations); set 4 (MY, BW^0.75, DIM, 33 WL, fat, protein, and lactose); set 5 (MY, BW^0.75, DIM, 33 WL, and visit duration in the feed bunk); set 6 (MY, DIM, and 33 WL); set 7 (MY, BW^0.75, and DIM); set-WL (included 361 WL); and set-BN (included just 33 selected WL). All models (i.e., each combination of covariate set and fitting approach, ANN or PLS) were validated with an external data set. The use of ANN improved the performance of models 2, 5, 6, and BN. The use of BN combined with ANN yielded the highest accuracy and precision. The addition of individual WL compared with milk components (set 2 vs. set 3) did not improve prediction quality when using PLS. However, when ANN was employed, the model prediction with the inclusion of 33 WL was improved over the model containing only milk components (set 2 vs. set 3; concordance correlation coefficient = 0.80 vs. 0.72; coefficient of determination = 0.67 vs. 0.53; root mean square error of prediction 2.36 vs. 2.81 kg/d). The use of ANN and the inclusion of a behavior parameter, set 5, resulted in the best predictions compared with all other models (coefficient of determination = 0.70, concordance correlation coefficient = 0.83, root mean square error of prediction = 2.15 kg/d). The addition of milk spectra information to models containing cow variables improved the accuracy and precision of DMI predictions in lactating dairy cows when ANN was used. The use of BN to select more informative WL improved the model prediction when combined with cow variables, with further improvement when combined with ANN.     

鲫鱼新鲜度近红外定量预测模型的建立

为实现鲫鱼新鲜度的快速测定,本文基于近红外漫反射光谱定量分析技术和化学计量学方法,采集了144个鲫鱼鱼肉样品在1000~1799 nm范围内的光谱数据,测定了鲫鱼样品的p H、TVB-N含量、TBA含量和K值四种新鲜度指标;在确定近红外光谱数据最佳预处理方法和适宜波段的基础上,分别采用偏最小二乘法、主成分分析和BP人工神经网络技术、偏最小二乘法和BP人工神经网络技术建立了鲫鱼新鲜度定量预测模型。结果表明,鲫鱼样品四种指标数据范围均较大,可满足建模要求。以p H为鲜度指标时,采用偏最小二乘法和BP人工神经网络技术建立的模型最好,其定标相关系数为0.9945;以TVB-N、TBA和K值为鲜度指标时,采用偏最小二乘法建立的模型最好,其定标相关系数分别为0.9857、0.9985和0.9952。建立的四种鲜度指标定量模型均具有较好的预测能力。     

Elman网络近红外光谱技术同时测定鲜乳中三种主成分含量

采用Elman神经网络(反馈神经网络,Recurrent Network)结合近红外光谱技术建立鲜乳中的脂肪、蛋白质、乳糖定量分析模型.用偏最小二乘法(Partial Least SqHales.PUS)将原始数据压缩主成分,取前3个主成分的14个吸收峰值输入Elman网络,网络中间层神经元个数为53.Elman网络模型对样品中3个组分含量的预测决定系数(R2)分别为:0.985、0.951、0.967,表明所建Elman网络预测模型能够较准确预测鲜乳中脂肪、蛋白质和乳糖的含量,从而为近红外光谱的多组分定量分析提供了新思路.     

On the use of milk composition measures to predict the energy balance of dairy cows

Milk composition varies with energy status and was proposed for measuring energy balance on-farm, but the accuracy of prediction using monthly samples is not high. With automated sampling and inline milk analysis, a much higher measurement frequency is possible, and thus improved accuracy of energy balance determination may be expected. Energy balance was evaluated using data in which milk composition was measured at each milking. Three breeds (Danish Holstein, Danish Red, and Jerseys) of cows (623 lactations from 299 cows) in parities 1, 2, and 3+ were used. Data were smoothed using a rolling local regression. Energy balance (EBal) was calculated from changes in body reserves (body weight and body condition score). The relationship between EBal and milk measures was quantified by partial least squares regression (PLS) using group means data. For each day in lactation, the within-breed and parity mean EBal and mean milk measures were used. Further PLS was done using the individual cow data. The initial PLS models included 25 combinations of milk measures allowing a range of nonlinear effects. These combinations were as follows: days in milk (DIM); DIM raised to the powers 2, 3, and 4; milk yield; fat content; protein content; lactose content; fat yield; protein yield; lactose yield; fat:protein ratio; fat:lactose ratio; protein:lactose ratio; and milk yield:lactose ratio, together with 10 “diff()” variables. These variables are the current minus the previous value of the milk measure in question. Using group means data, a very high proportion (96%) of the variability in EBal was explained by the PLS model. A reduced model with only 6 variables explained 94% of the variation in EBal. This model had a prediction error of 3.82 MJ/d; the 25-variable model had a prediction error of 3.11 MJ/d. When using individual rather than group means data, the PLS prediction error was 17.3 MJ/d. In conclusion, the mean Ebal of different parities of Holstein, Danish Red, and Jersey cows can be predicted throughout lactation using 1 common equation based on DIM, milk yield, milk fat, and milk protein measures.     

Differentiation of perirenal and omental fat quality of suckling lambs according to the rearing system from Fourier transforms mid-infrared spectra using partial least squares and artificial neural networks analysis

Fourier transform mid-infrared (FT-IR) spectroscopy was evaluated as a tool to discriminate between carcasses of suckling lambs according to the rearing system. Fat samples (39 perirenal and 67 omental) were collected from carcasses of lambs from up to three sheep dairy farms, reared on either ewes milk (EM) or milk replacer (MR). Fatty acid composition of the samples from each fat deposit was first analyzed and, when discriminant-partial least squares regression (PLS) was applied, a perfect discrimination between rearing systems could be established. Additionally, FT-IR spectra of fat samples were obtained and discriminant-PLS and artificial neural network (ANN) based analysis were applied to data sets, the latter using principal component analysis (PCA) or support vector machines (SVM) as processing procedure. Perirenal fat samples were perfectly discriminated from their FT-IR spectra. However, analysis of omental fat showed misclassification rates of 9–13%, with the ANN approach showing a higher discrimination power.     

应用傅立叶变换近红外光谱检测婴幼儿配方奶粉中的乳糖含量

利用傅立叶变换近红外光谱（FT-NIR）技术结合偏最小二乘法（PLS）对婴幼儿配方奶粉中的乳糖进行快速检测分析。搜集 94个不同产地、不同品牌的婴幼儿配方奶粉中乳糖的实验室数据,并采集婴幼儿配方奶粉的近红外光谱图,选择最优的光谱预处理方法,优化、验证和建立模型,并预测6个未知品牌的婴幼儿配方奶粉样品的乳糖含量。结果表明,不同阶段的婴幼儿配方奶粉有非常相似 的近红外特征图谱,图谱处理后的方差为95.423 5%,该模型测定的乳糖值与高效液相色谱（HPLC）法测定的乳糖值之间的平均相对误 差均≤0.67%,相对标准偏差（RSD）均≤0.88%,均符合误差范围。该方法可以无损、快速、高效地测定婴幼儿配方奶粉中的乳糖含量。     

Infrared milk analyzers: Milk urea nitrogen calibration

Our first objective was to redesign a modified 14-sample milk calibration sample set to obtain a well-distributed range of milk urea nitrogen (MUN) concentrations while maintaining orthogonality with variation in fat, protein, and lactose concentration. Our second objective was to determine the within- and between-laboratory variation in the enzymatic spectrophotometric method on the modified milk calibration samples and degree of uncertainty in MUN reference values, and then use the modified milk calibration samples to evaluate and improve the performance of mid-infrared partial least squares (PLS) models for prediction of MUN concentration in milk. Changes in the modified milk calibration sample formulation and manufacturing procedure were made to achieve the desired range of MUN concentrations. A spectrophotometric enzymatic reference method was used to determine MUN reference values, and the modified milk calibration samples were used to calibrate 3 mid-infrared milk analyzers. The within- and between-laboratory variation in the reference values for MUN were 0.43 and 0.77%, respectively, and the average expanded analytical uncertainty for the mean MUN value of the 14-sample calibration set was (mean ± SD) 16.15 mg/100 g ± 0.09 of milk. After slope and intercept adjustment to achieve a mean difference of zero with the calibration samples, it could be seen that the standard deviation of the differences of predicted versus reference MUN values among 3 different instruments and their PLS models were quite different. The orthogonal sample set was used (1) to determine when a PLS model did not correctly model out the background variation in fat, true protein, or anhydrous lactose; (2) to calculate an intercorrection factor to eliminate that effect, and (3) to improve the model performance (i.e., 50% reduction in standard deviation of the difference between instrument predictions and reference chemistry values for MUN).     

Sodium dl-α-tocopheryl phosphate inhibits ultraviolet-induced production of reactive nitrogen species in human keratinocytes

The phenomenon of aging can broadly be categorized into photoaging caused by exogenous factors and physiological aging that is caused by endogenous factors. Our goal was to develop a non-invasive way to assess changes taking place inside the skin for each type of aging, photoaging and physiological aging, by using near-infrared diffuse reflectance (NIR-DR) spectroscopy. For the photoaging and physiological aging effects, the outer forearm (sun-exposed) and the inner upper arm (sun-protected) skin areas were studied in eighty-six females from twenty-three to sixty-nine years of age. Measurements were made using NIR-DR and subjected to Principal Component Analysis (PCA); the results suggested the possibility of distinguishing and quantifying both types of aging taking place inside the skin by using the 1670–1820 nm and 2000–2230 nm regions of NIR-DR spectra. In photoaging, structural changes in proteins occur which are reflected in the NIR-DR spectra in the form of a peak shift near 2050 nm that is due to a combination of amide A and amide II. On the other hand, physiological aging is associated with a change in collagen quantity as is reflected in the portion of the NIR-DR spectra assigned to protein. Using NIR-DR and PCA, we discovered the possibility of using a non-invasive method for assessing the degree of photoaging and physiological aging as degeneration and degradation.
Key words:  photoaging, physiological aging, non-invasive, NIR-DR, PCA, peak shift, a combination of amide A and amide II, protein, degeneration, degradation.     

Comparison of the usability of different spectral ranges within the near ultraviolet, visible and near infrared ranges (UV-VIS-NIR) region for the determination of the content of scab-damaged component in blended samples of ground wheat.

Soft wheat grain samples of the same variety were obtained from a plot where the crop grew under natural conditions (control material) and from a plot where the crop was inoculated with Fusarium culmorum. The grain was ground and sieved with the finest fraction (a particle size less than 0.18 mm) of both materials being used for the preparation of samples in which the content of damaged constituent varied from zero to approximately 84%. Diffuse reflectance spectra of the absorbance from the blended samples were recorded in the 200-2500 nm spectral range and multivariate calibration PLS (Partial Least Squares) models were built within three spectral ranges: 200-2500, 200-1400 and 1400-2500 nm. Before modelling, several variants for spectra pre-processing were tried: multiple scatter correction, single and double differentiation, in all cases with and without centring. Single differentiation followed by centring was found to be the best method for spectra pre-processing in all spectral ranges. Very good calibration models were obtained for the whole and shorter wavelengths spectral ranges, allowing the detection of 1.50 and 0.76% of the content of scab-damaged constituent, respectively. Two-dimensional correlation spectroscopy applied to the set of spectra enabled the assignment of spectral bands and an analysis of changes in the chemical composition caused by scab damage. It was found that the content of protein and lipids increased with an increase of the scab-damaged constituent, whereas the content of moisture and starch decreased.     

Short communication: Suitability of fluorescence spectroscopy for characterization of commercial milk of different composition and origin

Thirteen milk brands comprising 76 pasteurized and UHT milk samples of various compositions (whole, reduced fat, skimmed, low lactose, and high protein) were obtained from local supermarkets, and milk samples manufactured in various countries were discriminated using front-face fluorescence spectroscopy (FFFS) coupled with chemometric tools. The emission spectra of Maillard reaction products and riboflavin (MRP/RF; 400 to 600 nm) and tryptophan (300 to 400 nm) were recorded using FFFS, and the excitation wavelengths were set at 360 nm for MRP/RF and 290 nm for tryptophan. Principal component analysis (PCA) was applied to analyze the normalized spectra. The PCA of spectral information from MRP/RF discriminated the milk samples originating in different countries, and PCA of spectral information from tryptophan discriminated the samples according to composition. The fluorescence spectral data were compared with liquid chromatography-mass spectrometry results for the glycation extent of the milk samples, and a positive association (R² = 0.84) was found between the degree of glycation of α-lactalbumin and the MRP/RF spectral data. This study demonstrates the ability and sensitivity of FFFS to rapidly discriminate and classify commercial milk with various compositions and processing conditions.     

基于近红外光谱的芝麻油酸价含量的预测

采用近红外光谱分析技术对芝麻油的酸价含量进行检测,避免了传统的化学方法缺陷,同时在不破坏样品的前提下极大地提高了检测效率。对39个芝麻油样本的酸价光谱图进行光谱预处理优化,并选择适当的光谱范围,采用偏最小二乘法(PLS)和BP神经网络算法进行了定量分析研究。结果表明,在所选定的样本和光谱范围内,PLS和BP神经网络算法均可以用于芝麻油酸价含量的预测,采用PLS模型的预测均方根误差(RMSEP)为0.058;用BP神经网络预测的RMSEP为0.148 8,偏最小二乘法建模相对于一般的BP网络建模方法更具有较好的建模预测效果。     

The Effect of Homogenisation and Storage on the Near-Infrared Spectra of Half Shell Pacific Oysters (<Emphasis Type="Italic">Crassostrea gigas</Emphasis>)

The effect of sample homogenisation and storage on the near-infrared spectra of Pacific Oysters (Crassostrea gigas) has been assessed. On each day of storage (Days 0, 3 and 5), spectra were collected using a Fourier transform near-infrared reflectance spectrometer in reflectance mode between 833 and 2,630 nm from whole (n = 20) and homogenised oysters (n = 20). The raw spectra were dominated by water- and fatty-acid-associated bands. Linear regression analysis of the water-associated absorbance bands occurring at 1,942 nm indicated that a physical or chemical interaction may be taking place within the oysters at or near Day 3, likely associated with transfer of liquids to and from oyster tissues. One-way analysis of variance of principal component scores and extended multiplicative scatter correction highlighted the water regions (O–H bonds) in whole oysters and the importance of N–H-related compounds in homogenised oysters throughout storage. These findings indicate the potential usefulness of near-infrared reflectance spectroscopy to monitor and evaluate degradation of oysters over time.     

基于不同波段近红外光谱的原料奶主要成分品质检测研究

试验采用不同波段的近红外光谱对原料奶的主要成分进行品质检测。使用2种近红外光谱仪采集原料奶的透反射和漫反射光谱,建立牛奶中蛋白含量、脂肪含量和乳糖含量的定量分析模型。结果表明,蛋白含量、脂肪含量、乳糖糖含量的相关系数(r)分别达到0.9311、0.9218、0.8288,预测误差均方根(RMSEP)分别为1.9144、2.0143、2.804,测量值与浓度参考值之间有着良好的相关性。结果表明,基于近红外光谱的原料奶主要成分品质快速检测准确度高,具有很高的实用价值。     

UV spectrophotometry method for the monitoring of galacto-oligosaccharides production

Monitoring the industrial production of galacto-oligosaccharides (GOS) requires a fast and accurate methodology able to quantify, in real time, the substrate level and the product yield. In this work, a simple, fast and inexpensive UV spectrophotometric method, together with partial least squares regression (PLS) and artificial neural networks (ANN), was applied to simultaneously estimate the products (GOS) and the substrate (lactose) concentrations in fermentation samples. The selected multiple models were trained and their prediction abilities evaluated by cross-validation and external validation being the results obtained compared with HPLC measurements. ANN models, generated from absorbance spectra data of the fermentation samples, gave, in general, the best performance being able to accurately and precisely predict lactose and total GOS levels, with standard error of prediction lower than 13 g kg⁻¹ and coefficient of determination for the external validation set of 0.93–0.94, showing residual predictive deviations higher than five, whereas lower precision was obtained with the multiple model generated with PLS. The results obtained show that UV spectrophotometry allowed an accurate and non-destructive determination of sugars in fermentation samples and could be used as a fast alternative method for monitoring GOS production.     

基于卷积神经网络的乳粉掺杂物拉曼光谱分类方法

提出一种基于卷积神经网络的乳粉掺杂物拉曼光谱分类方法。首先利用拉曼高光谱成像平台采集足量乳粉样品的原始光谱,然后利用离散小波变换对原始光谱进行预处理,将预处理后的光谱信号作为卷积神经网络输入构建模型,并分别比较光谱预处理前后的建模效果。结果表明,不合适的光谱预处理反而会降低卷积神经网络的分类效果,而原始拉曼光谱就能被卷积神经网络精准识别,所构建的原始光谱模型对实际未知样品的识别准确率为95.5%。结果表明,卷积神经网络具备光谱预处理与建模的一体化功能,可极大简化拉曼光谱分类识别的计算过程,对乳粉质量安全筛查具有重要意义。     

Simultaneous detection for adulterations of maltodextrin,sodium carbonate,and whey in raw milk using Raman spectroscopy and chemometrics

To achieve rapid on-site identification of raw milk adulteration and simultaneously quantify the levels of various adulterants, we combined Raman spectroscopy with chemometrics to detect 3 of the most common adulterants. Raw milk was artificially adulterated with maltodextrin (0.5–15.0%; wt/wt), sodium carbonate (10–100 mg/kg), or whey (1.0–20.0%; wt/wt). Partial least square discriminant analysis (PLS-DA) classification and a partial least square (PLS) regression model were established using Raman spectra of 144 samples, among which 108 samples were used for training and 36 were used for validation. A model with excellent performance was obtained by spectral preprocessing with first derivative, and variable selection optimization with variable importance in the projection. The classification accuracy of the PLS-DA model was 95.83% for maltodextrin, 100% for sodium carbonate, 95.84% for whey, and 92.25% for pure raw milk. The PLS model had a detection limit of 1.46% for maltodextrin, 4.38 mg/kg for sodium carbonate, and 2.64% for whey. These results suggested that Raman spectroscopy combined with PLS-DA and PLS model can rapidly and efficiently detect adulterants of maltodextrin, sodium carbonate, and whey in raw milk.     

不同品种原料乳理化特性分析

主要分析荷斯坦牛、牦牛、娟珊牛、摩拉水牛、尼里-拉菲水牛、Ⅰ代杂交水牛、高代杂交水牛等7个品种的原料乳的常规营养成分,并对原料乳中蛋白质和氨基酸组成及牛乳缓冲能力进行测定。结果显示：摩拉水牛、尼里-拉菲水牛、Ⅰ代杂交水牛和高代杂交水牛的乳脂肪含量分别为6.86%、7.99%、8.34%、8.69%,蛋白质含量分别为5.75%、5.14%、5.78%、5.58%,干物质含量分别为17.07%、18.79%、19.73%、19.88%,显著高于其他3种牛乳;牦牛和娟珊牛乳中乳糖含量分别为5.09%、5.17%,显著高于其他5种牛乳。SDS-PAGE显示：水牛乳中除含有牛乳血清蛋白(BSA)、α-酪蛋白(α-CN)、β-酪蛋白(β-CN)、κ-酪蛋白(κ-CN)、β-乳球蛋白(β-Lg)和α-乳白蛋白(α-La)主要蛋白外,还含有一些未定性蛋白;且水牛乳具有最好的缓冲性能,其次是牦牛乳和娟珊牛乳,荷斯坦牛乳缓冲性能最差。