Komposisi Mineral Keju Dengan Berbagai Waktu Pemasakan Dengan NIR

of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  The mineral composition (Ca, P, Mg, K, Na) in cheeses (cow’s, ewe’s and goat’s)with different ripening times using near infrared spectroscopy with a fibre-opticprobe I. González-Martín a, ⇑ , J.M. Hernández-Hierro a , I. Revilla b , A. Vivar-Quintana b , I. Lobos Ortega a a Departamento de Química Analítica, Nutrición y Bromatología, Facultad de Ciencias Químicas, C/Plaza de la Merced s/n, 37008 Salamanca, España, Spain b  Área de Tecnología de los Alimentos de la Universidad de Salamanca en Zamora, Spain a r t i c l e i n f o  Article history: Received 18 June 2009Received in revised form 25 October 2010Accepted 29 December 2010Available online 8 January 2011 Keywords: Mineral compositionCheesesNear infrared spectroscopyDetermination a b s t r a c t The results of this work show that it is possible to rapidly quantify calcium, phosphorus, magnesium,potassiumand sodiuminunknowncheeses elaborated withpercentages (0–100%) of milk fromdifferentspecies (cow, ewe, goat) by direct application of the fibre-optic probe on the sample without previousdestruction or treatment of the sample. Of the total number of samples, 170 were used to develop thecalibration models using the Modified Partial Least Squares (MPLS) regression method and 57 sampleswere used for external validation. The multiple correlation coefficients (RSQ) and prediction correctedstandard errors (SEP (C)) obtained for calcium (0.74, 0.64), phosphorus (0.69, 0.29), potassium (0.86,0.13), and sodium (0.92, 0.71) in g/kg respectively and magnesium (0.72, 30.9) in ppm, indicated thatthe models developed allow the determination of Ca, P, K, Na and Mg in unknown samples of cheesesof varying compositions up to 6months of ripening.   2011 Elsevier Ltd. All rights reserved. 1. Introduction Theconsumptionofcheeseisofgreatnutritionalinterestdueinparticular to its composition of micronutrients. The mineral con-tent of the milk depends on numerous factors, such as geneticcharacteristics, the stage of lactation, environmental conditions,type of pasture and soil contamination among others (Anguita,1996; Cichoscki, Valduga, Valduga, Tornadijo, & Fresno, 2002;Gambelli, Belloni, Ingrao, Pizzoferrato, & Santaroni, 1999; Lucaset al., 2006a, 2006b; Park, 2000; Patiño, Faisal, Cedres, Mendez, &Guanziroli-Stefani, 2005). The heavy metal in the cheese is funda-mentally related to the manufacturing practices and possible con-tamination from the equipment during the process (Mendil, 2006;Moreno-Rojas, Pozo-Lora, Zurera-Cosano, & Amaro-Lopez, 1994).The minerals like sodium, potassium, calcium and magnesiumand the anionic constituents, phosphate, citrate, chloride, carbon-ate and sulphate are found in the milk or associated to the caseinmicelle(Anguita,1996;Moreno-Rojasetal.,1994). Theirparticipa-tion has been demonstrated in the coagulation, drainage of wheyand the texture of the curds and properties such as the stabilityto heat and the capacity to coagulate depend on them (Patiñoet al., 2005). Particularly during cheese ripening some of the min-eral salts may migrate from the central part towards the externallayerofthecheeseblockorviceversaduetothepHgradienteffectcausing changes inthe concentrationof some elementsin the finalproduct (Moreno-Rojas et al., 1994).Theanalysisofthemineralcompositioninmilkandcheeseshasbeen performed by inductively coupled plasma atomic emissionspectrophotometry(Prieto, Franco, González, Bernardo, &Carballo,2002), ion exchange liquid chromatography, instrumental neutronactivation analysis (Gambelli et al., 1999), atomic absorptionspectrophotometry, atomic emission spectrophotometry (Lucas,Andueza, Rock, & Martin, 2008; Moreno-Rojas, Amaro-Lopez,Garcia-Gimeno, & Zurera-Cosano, 1995; Moreno-Rojas, Amaro-Lopez, & Zurera-Cosano, 1992) and flame and graphite furnaceatomic absorption spectrometry (Mendil, 2006).Inthelastdecadestechniqueslikenearinfra-red(NIR)spectros-copyhavebeendevelopedforthedeterminationofmajorityparam-eters in cheese (Rodríguez Otero, Hermida, & Centeno, 1997;Mazerolles, Duboz,&Hugot,2000), controlofripening,orsensorialcharacteristics (Ortiz, Sarabia, Garcia-Rey, & De Castro, 2006;Karoui, Pillonel, Schaller, Bosset, & De Baerdemaeker, 2007). Theprediction of the mineral composition in fresh and freeze-driedcheeses by NIR has been performed by (Lucas et al., 2008), previ-ously grinding the samples of cheese.Recently,theuseofNIRStechnologyemployingaremotereflec-tancefibre-opticprobehasbeenstudiedfortheanalysisoftheper-centage of milk (cow’s, ewe’s and goat’s) used in the elaboration of cheeses with different ripening times and also for the determina-tion of peptides or texture (Gonzalez-Martin et al., 2007; Revilla 0308-8146/$ - see front matter    2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.foodchem.2010.12.114 ⇑ Corresponding author. Tel./fax: +34 23 294483. E-mail address: (I. González-Martín).Food Chemistry 127 (2011) 147–152 Contents lists available at ScienceDirect Food Chemistry journal homepage:  et al., 2009; González-Martín, Hernández-Hierro, Vivar-Quintana,Revilla, & González-Pérez, 2009).In the present work we study the use of NIRS technology to-gether with the use of a remote reflectance fibre-optic probe (witha 5  5cm quartz window) for the determination of mineral com-position (Ca, P, Mg, K, Na) in 227 samples of cheeses elaboratedwith 16 varying mixtures of milk from cows, ewes and goats withdifferent ripening times over 6months. The NIR spectra were re-corded in reflectance mode, applying the fibre-optic probe directlyonto the cheese sample, with no sample preparation or manipula-tion. The reference mineral content was determined by ICP-opticspectroscopy. 2. Material and methods  2.1. Samples and cheese-making procedure To perform the present study a total of 227 cheeses of knowncomposition were elaborated and controlled (Gonzalez-Martinet al., 2007). Cheeses were prepared in the laboratory followingthis procedure: raw milk (40L), not standardised, was incubatedwith 15mgL   1 direct-vat-set starters made of   Streptococcus lactis,cremoris  and  Streptococcus diacetilactis , (MA400, Arroyo Laborato-ries, Santander, Spain) at 30  C. After 10min at 32  C, 12.5mgL   1 of calf rennet (90% chymosin, 10% trypsin, 1:150,000 strength)was added to each vat. Coagulation was allowed to take place over35–40min. When the curds had developed the desired firmness,evaluated subjectively, they were cut with a cheese harp untilpieces similar in size to a grain of rice (  6mm) were obtained.The curd was then stirred for 30min and heated for 10–20minat 37  C until it had reached the desired consistency to improveits drainage with sieves. The curd was packed in round hoops(1kg) and pressed for 6h at 1.5kgcm  2 at 20  C, with the inver-sion of the pieces of cheese at the mid-time of the pressed manu-facturing. After pressing, the cheeses were salted by soaking themin sodium chloride brine (18% w/v) at 18  C for 6h. The cheeseswere then moved to a drying chamber, where temperature(15  C) and relative humidity (70%) were controlled. They weremade of milk collected directly from farms in winter and summer.Bovine, ovine and caprine raw milk were obtained directly fromthe producers in Zamora (Spain). Cheeses with different composi-tions were elaborated, prepared with known, varying amounts of milk from cows, ewes and goats, with percentages ranging be-tween 0% and 100% (Gonzalez-Martin et al., 2007; Revilla et al.,2009). These were cylindrical, with an initial diameter of 10cmand a thickness of 5cm and they were monitored over 6months(at 0.2, 1, 2, 3, 4, 5 and 6months), using one of the pieces at eachtime. Each time, a cheese was monitored, the piece was cut in half transversally and a rectangular slice of approximately 2cm inthickness was collected.Ofallthe227samples,170wereemployedforthedenominatedcalibration set, and the other 57 were used for the externalvalidation.  2.2. Plasma ICP-optic spectroscopy The chemical measures of reference were performed by ICP-OES, after mineralisation in a system of microwaves Model EthosPlus (Milestone). The samples of cheese were ground and driedwith a heater at 105  C for 24h and subjected to mineralisationin a microwave system. About 0.5g of dry, ground sample waslater introduced into a high pressure capsule. In a first stage5mLof HNO 3  wereadded(c)andapotentialof300Wwasappliedfor 5min. Once the sample was cold a further 5mL of HNO 3  wereadded (c) and 1mL of H 2 O 2  30%, applying a potential of 300W for7min. The sample was cooled to room temperature, made up to100mL with distilled water and maintained at 4  C until analysis.A Jobin Yvon ICP OES model Ultima2 powered by a radiofre-quencygeneratorat1100Wwasusedforelementaldetermination.Thefollowingconditionswereusedtocarryouttheelementalanal-ysis: speed of the bomb: 20rpm; plasma argon gas flow PL1: 12L/min; pod gas flow G1: 0.2L/min; nebulizer flow: 1.0L/min; nebu-lizer pressure: 2.95bars; with the use of an argon humidificator.The analytical lines used for the different elements were Ca:317.9333nm; K: 766.490nm; Mg: 279.553nm; Na: 589.592nm;and P: 177.440nm. The range of the standards concentrations hasbeen used for Ca, Na and P: 5–10–20ppm; for K and Mg: 0.25–0.50–1ppm (concentrations of the standard dissolution which inthe equivalence in cheese were: Ca, Na and P: 1000–2000–4000mg/kg;andKandMg:100–200–400mg/kg).Inthosesampleswhichtheconcentrationwasnotfoundinitsrespectivelinealmar-gin were diluted with ultrapure water and were re-analysed untiltheywereinthelinealresponsemargin.Thedetectionlimitswere:Ca:10mg/kg;K:22mg/kg;Mg:10mg/kg;Na:36mg/kg;P:28mg/kg.The mineralisation and the ICP measurements were carriedout in the Servicio de Análisis Químico of the Universidad deSalamanca, Spain.  2.3. NIR spectroscopy AFoss NIRSystem5000 with a standard 1.5m210/7210 bundlefibre-optic probe, Ref. no. R6539-A, wasused. Theprobeemploys aremote reflectance system and uses a ceramic plate as reference.The window is of quartz with a 5  5cm surface area, measuringreflectance in the IR zone close to 1100–2000nm. The measure-ment of the spectra was carried out using NIRS technology and aremote reflectance fibre-optic probe that was applied directly tothe cheese samples with no prior treatment or manipulation. Thespectra were recorded at intervals of 2nm, performing 32 scansfor both the reference and samples. To minimise sampling error,all the samples were analysed in triplicate. The software usedwas Win ISI 1.50 installed on a Hewlett–Packard Pentium IIIcomputer.  2.4. Statistical analyses Calibrations were developed using WinISI II version 1.5 (Infra-soft International). The samples were divided into calibration( n =170)andvalidation( n  =57)sets.Alltypesofcheeseswererep-resented in both the validation and calibration sets. The modifiedpartial least squares (MPLS) regression method was used to obtainthe NIR equations for all the parameters studied (Shenk &Westhaus, 1995). Partial least squares (PLS) regression is similarto principal component regression (PCR), but uses both referencedata (chemical, physical, etc.) and spectral information to obtainthe factors useful for fitting purposes (Martens & Naes, 2001).MPLSisoftenmorestableandaccuratethanthestandardPLSalgo-rithm. In MPLS, the NIRS residuals at each wavelength obtainedafter each factor has been calculated are standardised (dividingbythestandarddeviationsoftheresidualsateachwavelength)be-fore calculating the next factor. When developing MPLS equations,cross-validation is recommended in order to select the optimalnumber of factors and to avoid overfitting. For cross-validation,the calibration set is partitioned into several groups. Each groupis then validated using a calibration developed on the other sam-ples. Finally, validation errors are combined into a standard errorof cross-validation (SECV) (Davies & Williams, 1996). It has beenreportedthat the SECV is the best single estimate of the predictioncapability of the equation and that this statistic is similar tothe average standard error of prediction (SEP) from 8 to 10 148  I. González-Martín et al./Food Chemistry 127 (2011) 147–152  randomly-chosen prediction sets. In all cases, cross-validation wasperformed by splitting the population into four groups. The effectsof scattering were removed using MSC (Multiplicative ScatterCorrection), SNV (Standard Normal Variate), DT (DeTrend) orSNV-DT, (Dhanoa, Lister, & Barnes, 1995). Moreover, the mathe-matical treatments were tested in the development of the NIRScalibrations. The following is an example: 1, 4, 4, 1, where the firstdigit is the number of the derivative, the second is the gap overwhich the derivative is calculated, the third is the number of datapoints in a running average or smoothing and the fourth is thesecond smoothing. The statistics used to select the best equationswereRSQ(multiplecorrelationcoefficients)andthestandarderrorof cross-validation (SECV). 3. Results  3.1. Chemical composition Table 1 shows the minimum, maximum and mean concentra-tions and standard deviations of the mineral composition in cal-cium, phosphorus, magnesium, potassium and sodium in thesamples of cheese studied obtained by the method of reference.The total of 227 samples was divided into two groups, calibration(170 samples) and external validation (57) sets and the Table 1presents the values for both groups.The average levels of calcium (8.1g/kg) are generally up to 10times higher in hard cheeses and 4 times more in cheeses withmould than in milk (Scott, 1989). The average concentration of phosphorus, 3.73g/kg, is in agreement with those found in curedcheeses of goat (3.60g/kg) (Almenara et al., 2007). The levels of calcium and phosphorus are higher in cheeses of cow in compari-sontothosefoundby(Raynal-Ljutovac, Lagriffoul, Paccard, Guillet,& Chilliard, 2008), whichshows that these two elements are foundinsimilarquantitiesinthemilkof goatandcow;theyarethemostabundant in ewe’s milk.The relationship Ca: P is 2.2, a high value compared with thoseobtained in goat’s cheeses (Park, 2000) and in cow’s cheeses( Jenness, 1980; Prieto et al., 2002). The average content of potas-sium, 1.26g/kg is in agreement with other works (Herrera et al.,2006). The quantity of potassium depends on the animal srcinof the milk, the highest values being presented in cow’s cheesesand the lowest in ewe’s. In addition, the concentrationof this min-eraldiminishesthroughouttheprocessofripening,whichsuggeststhat a considerable quantity is lost during the process of elabora-tionofthecheese,asaconsequenceoftheacidificationofthecurds(Almenara et al., 2007).The high concentration of sodium throughout the whole pro-cess of ripening, with an average value of 7.99g/kg, is principallydue to the addition of salt during the process of elaboration andthe period of ripening (Gambelli et al., 1999). Moreover, the sea-sonal variability is observed, thus the cow’s cheeses elaborated inwinter present lower concentrations of sodium than the cheeseselaborated from goat’s milk; the opposite of what occurs for thesamples elaborated in summer, the higher concentrations of so-dium are presented in cow’s cheeses in comparison to those of goat.The content of Mg does not present variations throughout theprocess of ripening and its average levels of 401.2ppm coincidewith those found in other studies (Cichoscki et al., 2002; Herreraet al., 2006; Moreno-Rojas et al., 1994).  3.2. Calibration equation To obtain the calibrations, a starting set of 170 cheese samplesof varying compositions of cow’s, ewe’s and goat’s milk was used.Fig. 1 shows the spectra of ten samples. Initially, a principal com-ponentanalysiswascarriedout(PCA).Inallcases,thespectralvar-iability explained was above 99%, and 12 principal componentswere required for calcium; 8 for phosphorus; 10 for potassium,10 for sodium and 11 for magnesium.The calibration process was implemented with the spectra of the resulting samples and their chemical data. The statisticalparametersofthecalibrationwereobtainedforeachofthecompo-nents after removing the samples for spectral (H criterion) orchemical reasons (T criterion). Anomalous spectra were detectedby applying the Mahalanobis distance. Furthermore, the risk of   Table 1 Main mineral composition of the samples of cheese. Calibration set ( N   =170) External validation set ( N   =57)Element Minimum Maximum Mean SD Minimum Maximum Mean SDCa (g/kg) 4.49 40.38 8.11 3.8 5.12 12.56 7.89 1.6P (g/kg) 2.43 6.03 3.73 0.7 2.35 5.68 3.83 0.8K (g/kg) 0.62 2.17 1.26 0.4 0.73 1.88 1.23 0.3Na (g/kg) 2.76 13.92 7.99 2.7 3.83 12.96 7.74 2.2Mg (ppm) 257.85 686.69 401.19 67.0 269.43 561.96 405.56 58.8 N  : number of samples; SD: standard deviation. 1200 1300 1400 1500 1600 1700 1800 1900 2000 wavelenghts, mn    L  o  g   (   1   /   R   ) (a) Calcium -0.08-0.06-0.04-0.0200.020.040.061100 1200 1300 1400 1500 1600 1700 1800 1900 2000 wavelengths, nm   n  o  n  e   /   2   º   d  e  r   i  v  a   t  e   (   L  o  g   (   1   /   R   )   ) (b) Fig. 1.  (a) NIR spectra of 10 samples of cheese and (b) mathematical treatments(none/2   derivate) in the spectra. I. González-Martín et al./Food Chemistry 127 (2011) 147–152  149  there being mistakes inthe equations under practical conditions isvery low or almost null when the standardised H statistic (Maha-lanobis distance) is used during routine analysis of unknown sam-ples. This tells us how different the spectrum of the unknownsample is from the average spectrum in the calibration set. Sam-ples with an H-value greater than three may be considered asnot belonging to the population from which the equations aredeveloped, and in this case the equations should not be used tomake any prediction; 7 samples were removed for calcium; 9 forphosphorus, 7 for potassium, 6 for sodium and 7 for magnesium.Calibrations were performed by modified partial least squaresregression (MPLS). Using the  T  P 2.5 criterion, samples that weredifferentfromthepopulationowingtothechemicalcriterionwereremoved from the set. On the basis of this chemical criterion, 7samples were removed for calcium, 8 for phosphorus, 5 for potas-sium, 14 for sodium and 8 for magnesium.  Table 2 Descriptors of NIR calibration. Element  N   Mathematical treatment SD Estimate RSQ SEC SECV  N   fac PLS Cross-validation groups RPDCa (g/kg) 156 None 2, 4, 4, 1 1.3 3.75–11.52 0.74 0.65 0.82 8 4 2.0P (g/kg) 153 Standard MSC 1, 4, 4, 1 0.5 2.03–5.28 0.69 0.30 0.33 8 4 1.8K (g/kg) 158 Standard MSC 2, 10, 10, 1 0.3 0.20–2.30 0.86 0.13 0.16 8 4 2.7Na (g/kg) 150 Standard MSC 2, 8, 6, 1 2.7 0–15.88 0.92 0.74 0.80 10 4 3.8Mg (ppm) 155 Detrend only 0, 0, 1, 1 60.4 216.17–578.82 0.72 31.95 40.25 10 4 2.0 N  : number of samples; MSC: Multiplicative Scatter Correction , SD: standard deviation; RSQ: multiple correlation coefficients; SEC: standard error of calibration;SECV: standard error of cross-validation; PLS: partial least squares; RPD: ratio performance deviation. 5.5 6.5 7.5 8.5 9.5 10.5 11.5 Calcium (g/kg).NIR    C  a   l  c   i  u  m    (  g   /   k  g   ) .   R   E   F CALCIUM RSQ: 0.76SEP: 0.64SEP(c): 0.64 2.533.544.555.562.5 3 3.5 4 4.5 5 5.5 Phosphorus (g/kg).NIR    P   h  o  s  p   h  o  r  u  s   (  g   /   k  g   ) .   R   E   F PHOSPHORUS RSQ: 0.70SEP: 0.29SEP(c): 0.29 7.5 12.5 17.5 Sodium (g/kg).NIR    S  o   d   i  u  m    (  g   /   k  g   ) .   R   E   F SODIUM RSQ: 0.93SEP: 0.71SEP(c): 0.71 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 Potassium (g/kg).NIR    P  o   t  a  s  s   i  u  m    (  g   /   k  g   ) .   R   E   F POTASSIUM RSQ: 0.87SEP: 0.13SEP(c): 0.13 250300350400450500550600650700250 300 350 400 450 500 550 600 Magnesium (ppm). NIR    M  a  g  n  e  s   i  u  m    (  p  p  m   ) .   R   E   F MAGNESIUM RSQ: 0.74SEP: 30.80SEP(c): 30.90 Fig. 2.  Correlation of the values obtained in the laboratory with respect to those predicted by NIR with a fibre-optic probe for the main mineral composition in cheese.150  I. González-Martín et al./Food Chemistry 127 (2011) 147–152
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks