ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 1073 FOOD COMPOSITION AND ADDITIVES Determintion of Ft, Moisture, nd Protein in Met nd Met Products y Using the FOSS FoodScn Ner-Infrred Spectrophotometer with FOSS Artificil Neurl Network Clirtion Model nd Associted Dtse: Collortive Study SHIRLEY ANDERSON 1 FOSS North Americ, 8091 Wllce Rd, Eden Pririe, MN 55344 Collortors: S. Aldn; M. Beggs; J. Birkey; A. Conquest; R. Conwy; T. Hemminger; J. Herrick; C. Hurley; C. Ionit; J. Longind; S. McMignl; A. Milu; T. Mitchell; K. Nnke; A. Perez; M. Phelps; J. Reitz; A. Slzr; T. Shinkle; M. Strmpe; K. Vn Horn; J. Willims; C. Wipperfurth; S. Zelten; S. Zerr A collortive study ws conducted to evlute the repetility nd reproduciility of the FOSS FoodScn ner-infrred spectrophotometer with rtificil neurl network clirtion model nd dtse for the determintion of ft, moisture, nd protein in met nd met products. Representtive smples were homogenized y grinding ccording to AOAC Officil Method 983.18. Approximtely 180 g ground smple ws plced in 140 mm round smple dish, nd the dish ws plced in the FoodScn. The opertor ID ws entered, the met product profile within the softwre ws selected, nd the scnning process ws initited y pressing the "strt" utton. Results were displyed for percent (g/100 g) ft, moisture, nd protein. Ten lind duplicte smples were sent to 15 collortors in the United Sttes. The within-lortory (repetility) reltive stndrd devition (RSD r ) rnged from 0.22 to 2.67% for ft, 0.23 to 0.92% for moisture, nd 0.35 to 2.13% for protein. The etween-lortories (reproduciility) reltive stndrd devition (RSD R ) rnged from 0.52 to 6.89% for ft, 0.39 to 1.55% for moisture, nd 0.54 to 5.23% for protein. The method is recommended for Officil First Action. Trditionl chemicl nlysis methods involve lengthy processes nd my tke severl hours to generte results, which in mny food production industries hinder rel-time production nd product-qulity decisions. Sumitted for puliction My 2007. The recommendtion ws pproved y the Methods Committee on Commodity Foods nd Commodity Products s First Action. See Officil Methods Progrm Actions, (2007) Inside Lortory Mngement, July/August issue. Corresponding uthor's e-mil: musch@fossnorthmeric.com 1 Decesed. Ner-infrred (NIR) technology offers rpid results, in 50 s, for multiple constituents nd is well suited to high-cpcity production environments. The met industry hs een routinely using NIR technology for mny yers for processing, pckging, nd leling. Currently, there re no Officil AOAC NIR methods for met nd/or met products. Therefore, it seemed pproprite tht this method should e evluted y collortive study, nd if the results were cceptle, tht the method should e mde n AOAC Officil Method for mets nd met products. This collortive study ws conducted to evlute the performnce of the FOSS FoodScn NIR spectrophotometer (Eden Pririe, MN) with rtificil neurl network (ANN) clirtion to support the officil use of the instrument within the met industries. The FoodScn nlysis is secondry method sed on NIR trnsmittnce technology for the simultneous determintion of moisture, protein, nd ft content in met nd met products. FoodScn Hrdwre From tungsten-hlogen lmp housed t the ck of the instrument, light is guided through n opticl fier into the internl moving-grting monochromtor, which provides monochromtic light in the spectrl region etween 850 nd 1050 nm. Through second opticl fier, light is then guided through collimtor lens positioned over the smple cup in the smple chmer. The light is trnsmitted through the smple, nd the unsored light strikes detector. The detector mesures the mount of light nd sends the result to the digitl signl processor, which communictes with the personl computer (PC) where the finl results re clculted. The smple is plced in cup nd positioned inside the FoodScn smple chmer. The smple cup is rotted during the nlysis process to suscn vrious zones of the test smple, which re then merged together for the finl result. This procedure provides more representtive result from potentilly nonhomogeneous smples.
1074 ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 Tle 1. Results of the ANN clirtion vlidtion Prmeter N SEP Men c Min. d Mx. e r 2f Ft 1403 0.78 21.10 0.30 70.60 0.9949 Moisture 1197 0.72 59.30 21.10 82.00 0.9968 Protein 1226 0.62 16.00 6.40 31.40 0.9744 N = Numer of smples in the independent vlidtion dt set. SEP = Accurcy expressed s stndrd error of prediction (SEP) in g/100 g. c Men = Averge vlue of the vlidtion set in g/100 g. d Min. = Minimum vlue in the vlidtion set in g/100 g. e Mx. = Mximum vlue in the vlidtion set in g/100 g. f r 2 = Coefficient of determintion etween FoodScn results nd chemicl nlysis results. Artificil Neurl Network Clirtion ANN clirtion is technique designed to emulte the sic function of the humn rin to solve complex prolems. The ANN model hs the ility to descrie oth liner nd nonliner reltionships etween spectrl chrcteristics nd compositionl nlysis. As demonstrted y Borggrd nd Thoderg (1, 2), ANN models hve een found to offer superior performnce, compred with other clirtion techniques such s prtil lest squres (PLS) regression models. Further, PLS clirtion models often cnnot dequtely descrie very rod rnge of smple types, physicl chrcteristics, nd composition within single clirtion. The ANN technique ws chosen for the FoodScn met clirtions ecuse of its ility to generte single, glol, multiproduct, full-rnge clirtion for ech constituent. Thus, ANN technology elimintes the need for development nd mintennce of seprte clirtions for specific smple types nd/or nlyte levels. ANN clirtions require dtse where ll clirtion smples re represented y spectr nd chemicl nlysis results for ech nlyte. Clirtion development uses chemometric techniques to correlte spectrl chrcteristics of the smple with compositionl nlysis. It is importnt tht clirtion smples descrie s much vriility s resonle nd prcticl. This vriility includes smple type, smple composition, nd results of chemicl nlysis nd will give the clirtion roustness with respect to these vriles. The FoodScn dtse contins over 21 000 met nd met product smples collected throughout the world. Constituent rnges of clirtion smples in the dtse re for ft, 0.1 86%; moisture, 10 81%; nd protein, 3 49%. The lortory procedures used for the ANN clirtions were officilly recognized methods for chemicl nlysis of met nd were predomintely Interntionl Stndrds Orgniztion (ISO) nd AOAC methods; however, ecuse smples cme from mny countries, mny loclly pproved methods were lso used. Methods included comustion nd Kjeldhl for protein, Soxhlet with nd without cid hydrolysis for ft, nd drying- nd vcuum-oven methods for moisture. The specific chemicl nlysis methods used in this study re identified in the Collortive Study section. The ANN clirtion is developed nd optimized y n itertive trining process on this dtse until clirtion errors hve een minimized. To ensure the clirtion hs not een overfitted, it is vlidted ginst n independent dt set representing smples nd reference lortories not included in the clirtion set. In routine nlysis, the test smple is expressed s result through mthemticl conversion of the spectrl dt y using the ANN clirtion model. Precollortive Clirtion Vlidtion Before the collortive study ws conducted, the ccurcy of the FOSS FoodScn nd ANN clirtion ws evluted. By using independent vlidtion dt sets (i.e., smples tht were not prt of the clirtion dt set), the results predicted y FoodScn were compred with results from lortory methods of chemicl nlysis. The vlidtion sets comprised etween 1200 nd 1400 smples from severl countries. They included vrious smple types (eef, poultry, pork, fresh, in-process, nd finished products) nd multiple instruments, plnt loctions, nd lortories providing chemicl nlysis vlues. The results of the vlidtion study re summrized in Tle 1. The ccurcy of the ANN clirtion is quntified in terms of stndrd error of prediction (or performnce; SEP) nd is clculted s defined y Mrk nd Workmn (3): SEP = ( (Y i i ) 2 /N) ½ where Y i = chemicl nlysis vlue of the ith smple; i = predicted vlue of the ith smple otined from the clirtion [ech smple in the vlidtion set will hve Y (chemicl nlysis) nd corresponding (NIR) vlue]; nd N = numer of smples in the vlidtion set. In routine usge, n SEP is clculted nd compred with the chemicl nlysis lortory error for ech constituent. As descried y Mrk nd Workmn (4), the SEP of good clirtion is generlly 1.0 1.5 times the lortory error s determined y lind duplicte study from minimum of 10 smples.
ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 1075 Figure 1. Correltion of chemicl nlysis vlues with FoodScn predicted vlues. The SEPs s shown in Tle 1 re in terms of constituent percent (g/100 g) nd were clculted from the vlidtion set s 0.62 for protein, 0.78 for ft, nd 0.72 for moisture. The coefficient of determintion (r 2 ) vlues from Tle 1 quntify the proportion of the smple vrince explined y the clirtion nd were found to e 0.9949 for ft, 0.9968 for moisture, nd 0.9744 for protein. These vlues were found to e cceptle on the sis of the criteri suggested y Willims (5) for NIR pplictions where r 2 vlues of >0.96 re deemed usle in most pplictions, including qulity ssurnce. The grphs in Figure 1 plot chemicl nlysis vlues (x) versus the FoodScn predicted vlues (y) for moisture, ft, nd protein for the vlidtion sets. Collortive Study The Study Director nd 15 met-plnt lortories prticipted in the collortive study. Lortories were selected to e representtive of those tht would routinely use the proposed method nd on the sis of vilility of the necessry instrumenttion. Prctice smples for testing were sent to ech collortor efore the ctul study smples, nd results were reported to the Study Director. This ws done to reherse ll spects of the study including smple preprtion, trnsporttion, hndling, nd ctul nlysis protocol s well s to determine the collortor s cpility to perform the study. In ddition, it ensured tht ll collortors' instruments contined the sme instrument settings, version of operting softwre, nd ANN clirtion, nd tht ll slope nd intercept (is) djustment options were set to 1 nd 0, respectively. The configurtion of instrument prmeters is identified in the FoodScn softwre s Profile, which ws identified s AOAC nd used for ll prctice nd study smples. The study smples were chosen to represent the mjority of products from the commercil met industry (eef, pork, nd poultry) nd included rw mets, emulsions, nd finished products. All smples were nturl nd rel-world, nd none were dulterted. The collortive study smples consisted of 10 met study smples prepred s lind duplicte pirs, resulting in 20 test smples. The study smples were ground ccording to AOAC Officil Method 983.18 (6). Of these smples, 6 were fresh met, 2 were emulsions, nd 2 were finished (cured) products. The use of fresh met smples presented severl logisticl nd prcticl chllenges. Becuse of shipping nd import/customs restrictions, prticipnts in the study were required to e U.S. sed. All study smples hd to e held t refrigertor temperture (2 4 C) to void deteriortion nd/or wter purge. Smples were not to e frozen. Smples were cold pcked nd shipped vi overnight crriers to the collortors, who were notified to expect delivery nd instructed to plce them under refrigertion nd conduct the nlyses within 24 h. There ws 10-dy turnround for the reference lortory conducting the chemicl nlyses of the study smples. Becuse the smples could not e withheld from the collortors until the results of the nlyses y the reference lortory were otined, they were shipped to the reference lortory nd the collortors simultneously. Another chllenge ws to void the plethor of issues ssocited with interspecies contmintion nd hygiene from the presence of nonproduction mets in the met plnts. These prolems were overcome y sending the smples to the collortors preground nd seled in the smple cups. After nlyzing the smples, the collortors returned them to the Study Director in iohzrd gs. For severl collortors, permission from the onsite U.S. Deprtment of Agriculture (USDA) inspector ws required for them to prticipte in the collortive study, nd only if this smple mngement protocol ws oserved. The smples were evluted for homogeneity nd to determine chemicl nlysis reference vlues y using n independent nlyticl lortory (Covnce Lortories, Inc., Mdison, WI). Covnce is ccredited y the USDA Food Sfety nd Inspection Service in food chemistry for met nd poultry products. To determine homogeneity, 10 smple liquots for ech constituent were tken in rndom mnner from pproximtely 5 l ground smple mteril. Ten replicte nlyses for ech smple nd constituent were performed.
1076 ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 Tle 2. Results otined y the reference lortory for the determintion of homogeneity, ft, moisture, nd protein Protein Ft Moisture Smple Type Men, g/100 g s r CV, % Men, g/100 g s r CV, % Men, g/100 g s r CV, % Beef, low ft Fresh, ground 17.80 0.23 1.29 16.79 0.19 1.13 65.23 0.65 1.00 Beef, high ft 15.64 0.31 1.98 29.30 0.57 1.95 54.42 0.39 0.72 Pork, low ft 17.17 0.37 2.16 22.25 0.36 1.62 61.17 0.22 0.36 Pork, high ft 14.68 0.44 3.00 31.92 0.37 1.16 53.98 0.44 0.30 Chicken rest Fresh, whole muscle 22.36 0.15 0.67 3.27 0.19 5.81 73.75 0.40 0.54 Turkey rest 24.47 0.12 0.49 1.48 0.14 9.46 73.86 0.33 0.45 Pepperoni Finished product 20.50 0.60 2.93 41.86 1.17 2.80 28.91 0.83 2.87 Hrd slmi 20.41 0.37 1.81 33.82 0.56 1.66 39.63 0.42 1.06 Hot dog Emulsion 16.42 0.15 0.91 15.39 0.13 0.84 63.29 0.32 0.51 Summer susge 17.31 0.21 1.21 8.07 0.10 1.24 68.51 0.26 0.38 Ten replicte determintions. AOAC Officil Methods for met nlysis were used for these determintions: protein, comustion method, 992.15 (7); ft, Soxhlet method, 960.39 (8); nd moisture, 125 C ir dry, 950.46B() (9). For ech smple nd constituent, the men chemicl nlysis vlue, repetility stndrd devition (s r ), nd coefficient of vrition (CV) were otined (Tle 2). Smple constituent vlues rnged from 1.48 to 41.86% for ft, from 28.91 to 73.86% for moisture, nd from 14.68 to 24.47% for protein. As suggested y Thiex et l. (10), n cceptility criterion of 2.0% for the CV ws pplied to evlute homogeneity. Severl oservtions were noted during this study. The comprtively high CVs for ft in chicken (5.81%) nd turkey (9.46%) ppered to e due to low men results for ft. The s r vlues of 0.19 nd 0.14 for ft in chicken nd turkey, respectively, compre stisfctorily with the Soxhlet results from the AOAC collortive study s reported y Foster nd Gonzles (11). Becuse the CVs for the other constituents in these smples were under the 2% limit, these smples were considered homogeneous. In generl, the CVs for protein were somewht high, specificlly the CVs for the high- nd low-ft pork nd the pepperoni, which were ove the 2% criterion. This ws likely due to the 0.3 g smple size tht the reference lortory used for protein determintion, which ws not discovered until fter the chemicl nlyses were performed. For met smples, including most products in this study, 0.3 g should not e considered representtive smple size for the chemicl determintion of protein. When the CVs for moisture in the pork smples were considered, these smples were lso considered homogeneous. Pepperoni ws the only smple for which the CVs for ll 3 constituents were >2%. The collortors were sked to report ll results on n s is sis, to perform only 1 nlysis per smple, nd to report results for ech constituent to 2 deciml plces. Two of the 15 collortors hd 2 instruments t their fcilities nd nlyzed the study smples on ech instrument; these dditionl nlyses resulted in totl of 17 dt sets. Study smple mterils were prepred nd shipped to collortors on July 17, 2006. Collortors were sked to nlyze the smples within 24 h fter receipt of the smples nd to hold them, if necessry, in the refrigertor. Study smple results were received from the collortors within 72 h. AOAC Officil Method 2007.04 Ft, Moisture, nd Protein in Met nd Met Products FOSS FoodScn Ner-Infrred (NIR) Spectrophotometer with FOSS Artificil Neurl Network (ANN) Clirtion Model nd Associted Dtse First Action 2007 [Applicle to the simultneous determintion of ft, moisture, nd protein in met nd met products (fresh met, eef, pork, nd poultry, emulsions, nd finished products) in the constituent rnges of 1 43% ft, 27 74% moisture, nd 14 25% protein.] Cution: (1) The FoodScn is designed to operte sfely under the following conditions: indoor use only; ltitude, 2000 m; temperture, 5 40 C; mximum reltive humidity, 80% for tempertures of 31 C, decresing linerly to 50% reltive humidity t 40 C; (2) ecuse of electric shock hzrd, covers or pnels must e removed y qulified personnel only; (3) the instrument is equipped with n AC power connector contining protective ground ginst electric shock hzrd; (4) if the equipment is used in mnner not specified in the documenttion, the protection provided y the equipment my e
ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 1077 Tle 2007.04. Interlortory study results for the determintion of ft, moisture, nd protein in met nd met products Smple Constituent determined No. of ls Men, % s r r c RSD r, %d s R e R f RSD R, %g HorRt Rec., % Beef, low ft Ft 17 17.46 0.28 0.78 1.60 0.33 0.92 1.87 0.72 104.17 Moisture 17 62.30 0.41 1.15 0.66 0.64 1.79 1.02 0.48 95.51 Protein 17 18.92 0.33 0.92 1.76 0.64 1.79 3.36 1.32 106.68 Beef, high ft Ft 17 29.99 0.26 0.73 0.87 0.29 0.81 0.95 0.40 102.37 Moisture 17 51.96 0.48 1.34 0.92 0.66 1.85 1.26 0.57 95.48 Protein 17 16.13 0.22 0.62 1.37 0.40 1.12 2.51 0.95 103.10 Pork, low ft Ft 17 21.99 0.16 0.45 0.73 0.17 0.48 0.75 0.30 98.81 Moisture 17 60.51 0.20 0.56 0.33 0.33 0.92 0.55 0.25 98.92 Protein 17 16.71 0.20 0.56 1.18 0.44 1.23 2.61 1.00 97.40 Pork, high ft Ft 17 31.32 0.07 0.20 0.22 0.17 0.48 0.55 0.23 98.11 Moisture 17 53.29 0.22 0.62 0.40 0.38 1.06 0.71 0.32 98.73 Protein 17 14.53 0.13 0.36 0.88 0.35 0.98 2.44 0.91 99.00 Chicken rest Ft 17 3.25 0.07 0.20 2.10 0.14 0.39 4.08 1.22 102.44 Moisture 17 73.48 0.19 0.53 0.26 0.32 0.90 0.43 0.21 99.63 Protein 17 22.74 0.09 0.25 0.40 0.18 0.50 0.78 0.31 101.68 Turkey rest Ft 17 1.89 0.05 0.14 2.67 0.13 0.36 6.89 1.90 127.84 Moisture 17 73.69 0.17 0.48 0.23 0.29 0.81 0.39 0.19 99.78 Protein 16(1) 24.86 0.18 0.50 0.71 0.22 0.62 0.90 0.36 101.60 Pepperoni Ft 17 43.42 0.15 0.42 0.35 0.22 0.62 0.52 0.23 103.72 Moisture 17 27.29 0.20 0.56 0.74 0.42 1.18 1.55 0.64 94.40 Protein 15(2) 20.87 0.14 0.39 0.69 0.33 0.92 1.56 0.62 101.80 Hrd slmi Ft 17 32.22 0.16 0.45 0.50 0.29 0.81 0.89 0.37 95.26 Moisture 16(1) 38.81 0.13 0.36 0.35 0.16 0.45 0.41 0.18 97.93 Protein 16(1) 19.66 0.07 0.20 0.35 0.11 0.31 0.54 0.21 96.31 Hot dog Ft 17 15.05 0.14 0.39 0.92 0.25 0.70 1.64 0.62 97.82 Moisture 17 62.17 0.20 0.56 0.32 0.30 0.84 0.48 0.22 98.23 Protein 17 15.25 0.26 0.73 1.69 0.52 1.46 3.42 1.29 92.86 Summer susge Ft 17 7.79 0.11 0.31 1.43 0.44 1.23 5.60 1.91 96.53 Moisture 17 68.48 0.56 1.57 0.82 0.72 2.02 1.06 0.50 99.95 Protein 17 15.03 0.32 0.90 2.13 0.79 2.21 5.23 1.97 86.82 Numer of lortories retined fter removl of the numer of lortories in prentheses. s r = Repetility stndrd devition. c r = Repetility; r = 2.8 s r. d RSD r = Repetility reltive stndrd devition. e s R = Reproduciility stndrd devition. f R = Reproduciility; R = 2.8 s R. g RSD R = Reproduciility reltive stndrd devition.
1078 ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 impired; (5) do not touch identified hot surfces; (6) when lifting or moving the instrument, use proper lifting equipment. Results of interlortory study. Vlues for within-lortory [repetility reltive stndrd devition (RSD r )]: 0.22 2.67% for ft, 0.23 0.92% for moisture, nd 0.35 2.13% for protein; vlues for etween-lortories [reproduciility reltive stndrd devition (RSD R )]: 0.52 6.89% for ft, 0.39 1.55% for moisture, nd 0.54 5.23% for protein; HorRt: 0.23 1.91 for ft, 0.18 0.64 for moisture, nd 0.21 1.97 for protein. See Tle 2007.04 for the results of the interlortory study supporting cceptnce of the method. A. Principle The method uses the FOSS FoodScn (FOSS North Americ, 8091 Wllce Rd, Eden Pririe, MN 55344, USA) with rtificil neurl network (ANN) clirtion nd ssocited dtse. The method is sed on ner-infrred (NIR) trnsmission spectroscopy, secondry, correltive technique to predict the concentrtion of vrious constituents in iologicl or orgnic smples. The ground smple is plced in cup nd positioned inside the FoodScn smple chmer. The smple cup is rotted during the nlysis process to suscn 16 zones of the test smple, which re then merged together for the finl result. The ANN clirtion model is derived from dtse of smple spectr nd ssocited chemicl nlysis vlues. The ANN clirtion quntifies the reltionship etween the spectrl chrcteristics nd the constituent vlues to interpret the test spectr nd return the results for protein, ft, nd moisture. B. Apprtus () FOSS FoodScn system. NIR trnsmission, with moving grting monochromtor scnning the region from 850 to 1050 nm. () FoodScn ANN clirtion for met nd met products, version 3.00, with the ssocited dtse. The FoodScn for met comes complete with the operting softwre, ANN clirtion, nd required ccessories. (c) Polysulfone or glss-ottom smple cups. 140 mm (dimeter) 17.5 mm (height). Becuse of opticl vritions, the use of polystyrene dishes, such s Petri dishes, is not recommended. (d) Personl computer (PC). With the following minimum specifictions: XP (SP2) operting system, Intel Celeron or Pentium 4 processor, 2.8 GHz, 512 M RAM, 40 MB hrd-drive spce, CD nd floppy drives, nd USB ports. Items () (d) re from FOSS Anlyticl (Slngerupgde 69, DK-3400 Hillerød, Denmrk) from FOSS North Americ (Tel: 1-952-974-9892, Fx: 1-952-974-9823, www.foss.dk). C. Preprtion of Anlyticl Smple Grind or homogenize representtive smple, using stndrdized protocol s descried in 983.18. Pck pproximtely 180 g smple into the FoodScn smple cup. Avoid ir pockets in the smple, nd pck the smple level with the top of the smple cup nd in consistent mnner. The optiml smple temperture is 10 20 C; however, if mesurements outside this rnge re needed, ensure tht the tempertures of the smples do not vry y more thn ±5 C nd/or condenstion on the collimtor lens. D. Determintion (1) Turn power on for the unit, llow unit to wrm up, nd perform self-test dignostics. (2) Select the pproprite opertor ID nd product profiles. The product profile must specify the use of the FoodScn ANN clirtion for met nd met products, version 3.00. (3) Plce prepred smple into the smple cup. (4) Plce the smple cup in the holder in the instrument. Ensure tht the smple cup engges the index pin in the holder. Close nd lock the door. (5) Strt the nlysis y pressing the Strt utton. (6) Enter smple ID nd/or smple description. (7) When nlysis is complete, remove smple from the instrument. (8) Process nd/or record results. E. Clcultions The FoodScn softwre clcultes the results for ft, moisture, nd protein, which re displyed s percentges (g/100 g) to 2 deciml plces. F. Clirtion Vlidtion NIR nlysis is secondry, correltive technique, nd the results must e vlidted ginst those otined y chemicl nlysis methods. It is importnt to use chemicl nlysis methods tht re officilly pproved nd stndrdized such s AOAC Officil Methods for Protein (992.15), Ft (960.39), nd Moisture [950.46B()]. The purpose of vlidtion is to determine the degree of greement of the FoodScn results with those from chemicl nlyses, sed on nlysis of r 2 vlues, stndrd error of prediction (SEP) vlues, nd is sttistics. Clculte the coefficient of determintion (r 2 ) etween chemicl nlysis methods nd FoodScn predictions from minimum of 10 smples. Compre this vlue to generl clirtion guidelines (>0.96) nd/or historicl performnce. The SEP is compred with the lortory error for ech constituent nd is clculted s follows: ( (Y i i ) 2 /N) ½ where Y i = chemicl nlysis vlue of the ith smple; i = predicted vlue of the ith smple otined from the clirtion [ech smple in the vlidtion set will hve Y (chemicl nlysis) vlue nd corresponding (NIR) vlue]; nd N = numer of smples in the vlidtion set. A stisfctory SEP is 1.0 1.5 times the lortory error s determined y lind duplicte study.
ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 1079 Determine the is from the difference etween the men of chemicl nlyses nd the FoodScn results from minimum of 10 smples (men lortory men FoodScn). Evlute the is for significnce y determining the ±is limit, using the following formul: (3 S)/ N where S = stndrd devition of the differences etween the FoodScn prediction nd the chemicl nlysis result in the vlidtion set; N = numer of smples in the vlidtion set (minimum of 10). When this limit is exceeded, is (or offset) is implemented in the FoodScn softwre, nd the vlidtion process is then repeted. Reference: J. AOAC Int. 90, 1073(2007). Results nd Discussion The results otined from the collortive study re shown in Tle 3. In the first nlysis of the collortive dt, lortory rnking ws otined y using the procedure descried y Youden nd Steiner (12) to detect ny is mong collortors' instruments. Only protein for Lortory 3 ws found to hve smll ut significnt is. However, ecuse none of the other constituents from this lortory demonstrted is (in which cse n instrument issue my hve een indicted), this lortory ws included in the study. Further nlysis of the collortive study dt found 4 pirs of results tht were identified s outliers y either the Cochrn or the Grus test. All sttistics nd outlier detection, except lortory is, were determined y using worksheets otined through AOAC INTERNATIONAL for lind duplicte collortive study design (13). The sttisticl nlysis of the collortive study results is summrized in Tle 2007.04. The within-lortory RSD r rnged from 0.22 to 2.67% for ft, from 0.23 to 0.92% for moisture, nd from 0.35 to 2.13% for protein. The etween-lortories RSD R rnged from 0.52 to 6.89% for ft, from 0.39 to 1.55% for moisture, nd from 0.54 to 5.23% for protein. HorRt vlues, s descried y Horwitz (14), were clculted nd found to e cceptle, rnging from 0.23 to 1.91 for ft, from 0.18 to 0.64 for moisture, nd from 0.21 to 1.97 for protein. Three HorRt vlues were found to pproch the cceptility limit of 2: ft (1.91) nd protein (1.97) in summer susge nd ft (1.90) in turkey rest. For summer susge, these results hve een due to possile homogeneity, ingredient, or other opticl chrcteristics. The high HorRt vlue for ft in turkey ws cused y the very low men vlue. The pepperoni smple, which demonstrted high CVs of 2.93% for protein, 2.80% for ft, nd 2.87% for moisture in the homogeneity study, yielded very good HorRt vlues of 0.62, 0.23, nd 0.64, respectively. These results could e due to the FoodScn s use of lrger smple size thn ws used in the chemicl nlyses nd/or to the higher men vlues, compred with (for exmple) those for protein nd ft in the summer susge. Percent recovery ws determined y compring the overll men of the collortor vlues for ech study smple constituent with its chemicl nlysis vlue. In most cses, the recovery ws very good, etween 94 nd 107%. Three results demonstrted either high or low recovery: ft in turkey (128%), protein in hot dog (93%), nd protein in summer susge (87%). Bis or inherent systemtic error, s descried y Youden nd Steiner (15), is exhiited when the predicted results for specific smple group or product show men offset vlue when compred with their reference vlues. The is, i.e., difference etween the men of chemicl nlyses nd the FoodScn results from minimum of 10 smples (men lortory men FoodScn), my or my not e sttisticlly significnt. Bsed on the procedure descried in AOAC Officil Method 972.16 (16), the ±is limit cn e clculted y using the following formul: (3 S)/ N where S = stndrd devition of the differences etween the FoodScn prediction nd the chemicl nlysis in the vlidtion set; N = numer of smples in the vlidtion set (minimum of 10). When this limit is exceeded, is (or offset) is implemented in the FoodScn softwre, nd the vlidtion process is repeted. Refer to the Foss FoodScn User Mnul (17) for this procedure. Comprison of FoodScn nd Chemicl Anlysis Precision The precision vlues of the FoodScn study results were compred with those of officil chemicl nlysis methods for oth the study smples nd prior AOAC nd ISO collortive studies. Becuse AOAC Officil Method 950.46B() for moisture in met includes no precision dt, ISO 1442 (18) ws used. This method, sed on drying of the smple t 102 105 C for 16 18 h, hd een collortively studied in the erly 1990s, nd precision sttistics were determined. Tle 4 compres the precision prmeters found for the study smple chemicl nlysis nd collortive studies prior to the FoodScn collortive study. For protein nd moisture, the FoodScn collortive study dt demonstrted similr repetility nd reproduciility compred with the corresponding dt for chemicl nlysis methods. For ft determintions, the FoodScn demonstrted etter repetility nd reproduciility. Collortors' Comments There were reltively few collortor comments. Collortor 4 reported tht when the smples rrived, their temperture ws found to e 59 F (15 C), nd the following ws noted: Smple 4: Top of dish hs red-colored met ;
1080 ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 Tle 3. Interlortory study results for the determintion of ft, moisture, nd protein in 20 test smples prepred s 10 lind duplicte pirs B eef, high ft B eef, low f t C hicke n Hrd slm i Hot dog, e mulsion P epperon i P ork, high f t Pork, low f t Summer susge, e mulsion Turke y L 2 0 4 5 1 7 8 1 6 2 6 1 1 0 1 1 7 1 3 1 9 1 2 3 9 1 5 1 4 1 8 Ft, g/100 g A 3 0.3 0 3 0.1 1 1 7.6 0 1 7.5 3 3.1 8 3.1 3 3 2.2 0 3 2.0 9 1 4.9 6 1 4.9 5 4 3.2 2 4 3.2 3 3 1.4 9 3 1.3 3 2 1.8 8 2 1.9 4 7.9 9 7.7 2 1.7 1 1.7 1 B 3 0.2 2 3 0.1 9 1 7.6 9 1 7.2 1 3.2 7 3.2 9 3 2.0 7 3 2.0 5 1 4.9 3 1 4.9 2 4 3.1 7 4 3.4 1 3 1.5 0 3 1.4 3 2 2.1 9 2 1.9 6 7.7 8 7.7 3 1.8 9 1.9 4 C 3 0.3 7 3 0.1 3 1 7.9 2 1 7.8 8 3.5 0 3.2 9 3 2.6 6 3 2.8 9 1 5.0 9 1 5.3 9 4 3.8 6 4 3.6 8 3 1.3 7 3 1.4 8 2 2.0 7 2 1.8 8 8.0 3 8.2 9 2.0 3 2.0 4 D 3 0.1 0 2 9.8 9 1 7.7 7 1 7.2 6 3.5 6 3.5 0 3 2.3 0 3 2.6 9 1 5.1 7 1 5.2 0 4 3.7 3 4 3.5 7 3 1.1 6 3 1.2 7 2 1.8 4 2 1.8 9 8.6 5 8.4 8 2.0 4 2.0 6 E 2 9.9 3 2 9.9 5 1 7.4 1 1 7.0 0 3.4 1 3.4 4 3 1.8 8 3 2.0 1 1 4.7 8 1 4.5 7 4 2.9 7 4 3.0 2 3 1.1 0 3 1.0 3 2 2.3 1 2 1.9 6 7.7 8 7.5 6 1.8 6 1.8 6 F 2 9.7 7 3 0.0 6 1 7.9 7 1 7.9 1 3.4 2 3.3 0 3 2.2 0 3 2.3 4 1 5.6 7 1 5.5 6 4 3.6 4 4 3.6 2 3 1.1 9 3 1.3 0 2 1.9 6 2 2.0 5 7.2 9 7.1 1 1.9 1 1.9 8 G 2 9.5 0 2 9.8 7 1 7.6 4 1 7.4 3 3.2 6 3.3 2 3 1.7 7 3 2.0 2 1 4.8 1 1 4.9 9 4 3.2 8 4 3.3 7 3 1.2 9 3 1.0 8 2 1.9 9 2 1.7 8 7.9 4 7.9 5 1.7 3 1.7 3 H 3 0.1 3 3 0.0 9 1 7.6 9 1 7.2 1 3.2 1 3.3 4 3 2.2 7 3 2.2 2 1 4.6 5 1 4.7 6 4 3.6 9 4 3.3 8 3 1.1 4 3 1.1 5 2 2.0 7 2 1.6 3 7.9 1 7.8 8 1.8 9 1.8 4 I 3 0.2 7 2 9.4 3 1 7.6 5 1 7.0 1 3.4 5 3.2 7 3 2.0 3 3 2.4 3 1 5.1 9 1 5.2 5 4 3.4 2 4 3.4 3 3 1.0 1 3 1.0 7 2 1.7 5 2 1.9 7 7.0 2 7.0 7 1.9 9 1.9 4 J 3 0.2 2 3 0.4 5 1 7.9 7 1 8.1 2 3.1 7 3.2 8 3 1.9 9 3 1.9 9 1 5.3 4 1 5.0 6 4 3.6 1 4 3.8 1 3 1.5 3 3 1.6 0 2 2.4 6 2 2.0 0 7.1 7 7.0 4 1.6 7 1.7 7 K 2 9.6 1 2 9.6 3 1 6.9 7 1 7.0 5 3.5 7 3.7 0 3 2.1 1 3 2.0 6 1 5.3 0 1 5.1 5 4 3.2 7 4 3.3 1 3 1.5 8 3 1.6 0 2 2.2 5 2 2.0 6 7.6 8 7.7 6 2.2 5 2.1 2 L 3 0.0 4 3 0.5 6 1 7.0 1 1 7.7 9 3.0 9 3.1 6 3 2.1 1 3 2.4 1 1 5.3 0 1 4.9 6 4 3.2 1 4 3.4 7 3 1.5 0 3 1.4 7 2 1.8 7 2 2.0 9 7.2 3 7.3 6 1.6 9 1.8 7 M 2 9.6 5 3 0.1 7 1 7.5 7 1 7.2 1 3.4 9 3.3 9 3 1.9 0 3 2.0 2 1 5.1 4 1 4.9 2 4 3.4 6 4 3.2 6 3 1.3 9 3 1.4 7 2 2.0 6 2 2.2 0 7.7 7 7.8 6 1.9 3 1.8 7 M 2 9.9 8 3 0.1 4 1 7.5 1 1 7.1 7 3.2 8 3.3 4 3 2.8 3 3 2.5 4 1 5.1 9 1 5.0 2 4 3.5 7 4 3.4 4 3 1.2 8 3 1.2 7 2 1.8 7 2 1.8 6 8.0 0 8.1 0 1.7 8 1.8 7 N 3 0.0 5 2 9.8 2 1 7.4 8 1 7.4 9 3.3 3 3.3 1 3 2.3 2 3 1.9 5 1 4.9 7 1 4.9 3 4 3.0 9 4 3.5 4 3 1.3 6 3 1.2 7 2 1.8 9 2 1.9 4 8.0 3 7.8 1 1.8 5 1.8 4 N 3 0.0 3 2 9.3 4 1 7.2 7 1 7.6 2 3.4 1 3.3 7 3 2.1 4 3 1.8 5 1 5.0 1 1 4.6 9 4 3.0 7 4 3.5 0 3 1.2 5 3 1.1 4 2 2.0 5 2 1.9 7 8.1 0 7.9 2 1.9 1 1.8 9 O 2 9.7 5 3 0.0 5 1 7.0 7 1 7.6 1 3.4 2 3.4 4 3 2.4 7 3 2.5 6 1 4.8 9 1 5.1 2 4 3.4 4 4 3.4 6 3 1.2 8 3 1.3 4 2 2.0 1 2 1.8 0 8.5 0 8.3 4 1.9 1 1.9 6 Moisture, g/100 g A 5 2.1 0 5 2.0 5 6 2.0 8 6 1.8 5 7 3.4 1 7 3.3 6 3 8.8 2 3 8.6 8 6 2.6 0 6 1.9 7 2 7.5 5 2 7.4 6 5 3.2 4 5 3.2 1 6 0.4 0 6 0.2 0 6 9.4 8 6 8.5 8 7 3.4 9 73.4 7 B 5 1.2 0 5 1.3 6 6 2.0 8 6 2.4 6 7 3.5 4 7 3.3 9 3 9.0 2 3 8.9 0 6 2.4 4 6 2.3 1 2 7.4 4 2 7.3 1 5 3.5 7 5 3.1 0 6 0.3 3 6 0.7 6 6 9.5 3 6 9.0 9 7 3.6 7 73.8 0 C 5 1.7 5 5 2.3 3 6 2.4 8 6 2.5 6 7 3.3 5 7 3.7 6 3 8.9 0 3 8.6 6 6 1.9 8 6 1.9 6 2 7.6 9 2 7.7 5 5 3.8 5 5 3.8 0 6 0.8 4 6 1.0 4 6 7.3 0 6 7.6 5 7 3.9 6 73.6 8 D 5 2.0 3 5 1.9 9 6 2.4 7 6 2.9 1 7 3.8 4 7 3.7 0 3 9.0 4 3 8.8 2 6 1.9 7 6 1.9 4 2 7.7 7 2 7.7 8 5 3.4 5 5 3.5 6 6 0.5 5 6 0.9 4 6 7.9 9 6 9.2 3 7 4.1 1 74.0 4 E 5 1.3 0 5 2.3 1 6 1.7 8 6 2.3 5 7 3.6 9 7 3.5 0 3 8.7 6 3 8.6 9 6 2.3 2 6 2.8 2 2 7.1 5 2 7.1 2 5 2.9 7 5 3.1 1 6 0.0 1 5 9.9 8 6 7.2 4 6 9.2 3 7 3.6 1 73.8 1 F 5 2.2 7 5 1.5 8 6 1.6 3 6 2.0 6 7 3.7 7 7 3.5 0 3 8.8 0 3 8.6 4 6 2.2 5 6 2.2 9 2 6.8 3 2 6.9 3 5 3.7 7 5 3.3 4 6 0.9 0 6 0.9 7 6 7.6 7 6 7.5 5 7 3.7 2 73.8 4 G 5 3.0 4 5 2.4 0 6 2.2 3 6 2.5 3 7 3.3 7 7 3.2 7 3 9.1 1 3 8.9 7 6 2.7 4 6 2.3 7 2 7.7 2 2 7.7 0 5 3.5 0 5 3.7 9 6 0.6 0 6 0.5 8 6 8.8 0 6 9.2 0 7 3.5 1 73.6 6 H 5 2.3 3 5 2.0 5 6 1.9 2 6 2.8 9 7 3.4 6 7 3.4 9 3 8.8 1 3 8.7 3 6 2.3 8 6 2.2 6 2 7.1 4 2 7.4 0 5 3.3 7 5 3.7 9 6 0.0 5 6 0.7 3 6 8.4 7 6 7.6 6 7 3.3 4 73.5 6 I 5 1.1 6 5 2.0 7 6 2.1 9 6 3.3 0 7 3.4 5 7 3.8 8 37.2 2 38.63 6 2.44 6 2.4 1 2 6.4 1 2 6.8 2 5 3.3 1 5 2.9 3 6 0.6 0 6 0.3 1 6 8.1 0 6 9.1 1 7 3.8 0 73.9 4 J 5 0.8 9 5 0.4 5 6 0.6 0 6 0.4 2 7 2.9 7 7 2.7 3 3 8.8 3 3 8.6 6 6 1.8 9 6 2.0 9 2 6.9 7 2 7.2 0 5 2.6 5 5 2.6 8 6 0.1 2 6 0.1 6 6 9.2 8 6 9.3 7 7 3.0 7 73.3 4 K 5 2.7 9 5 3.4 1 6 2.6 9 6 2.9 8 7 3.7 7 7 3.9 4 3 9.0 0 3 8.9 5 6 2.1 5 6 2.2 0 2 6.4 3 2 6.4 6 5 3.5 5 5 3.2 2 5 9.8 2 6 0.2 4 6 8.1 4 6 7.7 1 7 4.4 0 74.1 2
ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 1081 T le 3. ( c ontinue d ) B eef, high ft B eef, low f t C hicke n Hrd slm i Hot dog, e mulsion P epperon i P ork, high f t Pork, low f t Summer susge, e mulsion Turke y L 2 0 4 5 1 7 8 1 6 2 6 1 1 0 1 1 7 1 3 1 9 1 2 3 9 1 5 1 4 1 8 L 5 2.2 5 5 1.8 6 6 2.7 3 6 1.3 3 7 3.0 7 7 2.8 4 3 8.8 6 3 8.3 6 6 1.5 5 6 2.0 6 2 7.6 9 2 7.7 9 5 3.2 1 5 3.4 4 6 0.5 6 6 0.3 3 6 7.7 2 6 8.4 2 7 3.5 3 73.0 9 M 5 2.0 1 5 1.3 5 6 2.4 4 6 2.5 4 7 3.5 6 7 2.8 6 3 8.8 9 3 8.7 5 6 2.2 6 6 2.1 4 2 7.0 3 2 6.9 5 5 2.9 4 5 3.0 7 6 0.6 2 6 0.5 3 6 7.8 4 6 8.4 9 7 3.4 1 73.6 6 M 5 2.4 9 5 1.9 9 6 2.9 4 6 2.9 8 7 3.3 5 7 3.4 3 3 8.8 1 3 8.5 7 6 1.9 6 6 1.7 8 2 7.5 1 2 6.9 5 5 3.8 3 5 3.7 8 6 0.8 8 6 0.9 6 6 7.6 5 6 8.4 9 7 3.7 6 73.7 3 N 5 1.1 0 5 2.1 9 6 2.4 1 6 2.3 2 7 3.4 1 7 3.4 0 3 8.6 6 3 8.6 6 6 2.0 8 6 2.0 1 2 7.6 4 2 6.9 9 5 3.0 2 5 3.3 8 6 0.7 5 6 0.7 8 6 8.7 9 6 9.4 0 7 3.6 5 73.4 6 N 5 0.9 1 5 2.3 2 6 2.2 9 6 2.0 7 7 3.7 5 7 4.0 0 3 8.9 4 3 9.0 2 6 2.2 3 6 2.6 1 2 7.7 6 2 7.2 7 5 2.9 4 5 2.2 9 6 0.7 7 6 0.4 6 6 9.4 0 6 9.2 0 7 4.0 4 73.5 8 O 5 2.6 3 5 2.6 7 6 3.1 8 6 2.4 9 7 3.7 4 7 3.7 4 3 8.8 9 3 8.7 5 6 1.6 1 6 1.7 7 2 7.7 7 2 7.5 2 5 3.1 1 5 3.1 6 6 0.2 0 6 0.3 6 6 8.0 7 6 8.4 1 7 3.7 7 73.9 9 Protein, g/100 g A 1 6.2 2 1 6.3 2 1 9.6 4 1 9.8 1 2 2.7 5 2 2.8 1 1 9.6 9 1 9.7 4 1 4.2 4 1 5.2 5 2 0.7 4 2 0.7 3 1 4.7 9 1 4.8 1 1 6.8 5 1 7.1 4 1 4.0 9 1 4.5 6 2 5.1 0 25.1 9 B 1 6.8 3 1 6.6 8 1 9.8 5 1 9.6 8 2 2.8 6 2 2.8 7 1 9.6 7 1 9.6 4 1 5.2 1 1 5.1 5 2 0.7 2 2 0.7 0 1 4.7 2 1 4.9 8 1 6.9 5 1 6.9 3 1 4.1 0 1 4.3 9 2 5.2 8 25.0 8 C 1 5.7 6 1 5.8 3 1 7.9 4 1 8.2 3 2 2.8 5 2 2.7 6 1 9.7 2 1 9.7 4 1 5.5 8 1 5.6 1 2 0.7 4 2 0.6 2 1 3.9 1 1 3.7 7 1 5.9 5 1 5.9 6 1 5.8 4 1 5.5 3 2 4.8 1 24.8 4 D 1 5.9 4 1 5.9 6 1 7.9 2 1 7.9 1 2 2.8 5 2 2.9 1 1 9.8 3 1 9.7 5 1 5.6 3 1 5.4 3 2 0.7 8 2 1.1 6 1 4.1 4 1 4.1 6 1 6.2 5 1 6.2 7 1 5.1 9 1 5.4 4 2 5.1 4 24.8 8 E 1 6.6 5 1 6.3 1 1 9.6 4 1 9.5 3 2 2.6 8 2 2.7 4 1 9.4 6 1 9.5 6 1 4.7 3 1 4.4 3 2 0.3 6 2 0.8 5 1 4.6 5 1 4.9 7 1 7.0 6 1 7.2 1 1 4.4 4 1 4.6 8 2 5.0 3 24.8 4 F 1 5.8 1 1 6.1 7 1 8.7 0 1 8.5 3 2 3.0 7 2 3.0 0 1 9.6 1 1 9.6 4 1 5.8 3 1 6.0 0 2 1.5 1 2 1.5 4 1 4.4 5 1 4.7 1 1 6.8 0 1 6.6 9 1 6.1 3 1 6.3 4 2 5.0 4 24.9 6 G 1 5.9 4 1 6.0 6 1 9.4 7 1 8.9 3 2 2.6 5 2 2.6 6 1 9.6 5 1 9.7 5 1 4.2 2 1 4.4 1 2 1.0 3 2 0.8 5 1 4.7 3 1 4.4 9 1 6.5 6 1 6.6 4 1 4.2 5 1 4.0 9 2 5.0 4 24.8 3 H 1 5.5 8 1 5.7 9 1 8.9 2 1 8.3 1 2 2.6 6 2 2.9 5 1 9.6 1 1 9.6 6 1 5.1 4 1 4.9 2 2 0.9 8 2 0.7 5 1 4.4 8 1 4.2 0 1 6.2 9 1 6.8 6 1 4.6 3 1 5.2 0 2 5.0 1 24.6 9 I 1 6.6 4 1 6.4 7 1 9.0 7 1 8.6 7 2 2.6 8 2 2.8 2 1 9.4 3 1 9.6 7 1 5.8 9 1 5.6 7 20.2 6 J 1 5.5 3 1 5.5 8 1 8.5 8 1 9.0 4 2 2.7 8 2 2.7 4 17.6 4 17.79 1 5.88 1 5.4 6 19.1 6 21.23 19.22 1 4.52 1 4.3 8 1 6.8 3 1 6.9 9 1 6.2 6 1 5.3 8 2 4.5 1 24.7 0 1 4.97 1 4.7 1 1 6.6 6 1 7.3 8 1 6.4 5 1 6.9 4 2 5.0 3 24.5 2 K 1 6.3 1 1 6.0 3 1 8.0 9 1 8.9 8 2 2.3 6 2 2.2 3 1 9.5 3 1 9.3 7 1 6.0 3 1 5.9 9 2 0.9 4 2 1.0 1 1 4.8 5 1 4.9 5 1 7.1 7 1 7.5 6 1 5.1 6 1 5.4 9 23.8 6 24.30 L 1 6.4 1 1 6.4 5 1 9.5 6 2 0.0 7 2 2.4 9 2 2.6 0 1 9.7 3 1 9.7 1 1 5.0 9 1 4.7 6 2 0.9 0 2 1.0 5 1 4.3 9 1 4.4 8 1 6.3 8 1 6.4 4 1 5.0 8 1 4.1 6 2 4.8 2 24.5 6 M 1 6.0 1 1 6.5 8 1 9.6 5 1 9.0 2 2 2.8 6 2 2.9 5 1 9.6 1 1 9.6 8 1 5.7 1 1 5.3 0 2 1.1 4 2 1.4 1 1 4.9 4 1 4.7 6 1 7.0 8 1 7.0 1 1 5.0 9 1 4.8 0 2 5.0 9 24.9 6 M 1 5.6 8 1 6.0 5 1 8.7 0 1 8.6 8 2 2.5 9 2 2.7 0 1 9.7 2 1 9.7 7 1 5.6 1 1 5.1 6 2 0.3 9 2 0.2 9 1 3.9 1 1 3.8 1 1 6.0 5 1 5.7 4 1 5.5 4 1 5.0 6 2 4.8 1 24.7 4 N 1 6.7 2 1 6.0 9 1 8.4 8 1 9.5 1 2 2.8 3 2 2.7 5 1 9.8 0 1 9.7 5 1 4.9 2 1 5.4 3 2 0.5 1 2 0.4 2 1 4.6 8 1 4.6 2 1 6.7 1 1 6.5 9 1 4.3 0 1 4.1 7 2 4.9 6 24.5 2 N 1 6.7 5 1 6.1 6 1 9.0 3 1 9.0 9 2 2.7 4 2 2.4 4 1 9.5 4 1 9.6 7 1 4.6 2 1 4.7 6 2 0.7 7 2 0.7 9 1 4.8 4 1 4.8 1 1 7.1 3 1 7.0 8 1 4.1 1 1 4.3 8 2 4.6 8 24.3 8 O 1 5.5 2 1 5.4 2 1 8.2 0 1 7.9 6 2 2.7 5 2 2.6 5 1 9.6 9 1 9.6 3 1 5.1 4 1 5.2 4 2 1.2 7 2 1.1 2 1 4.3 2 1 4.2 3 1 6.4 1 1 6.6 4 1 5.0 8 1 4.6 4 2 4.8 2 24.6 7 Excluded s n outlier y the Cochrn test. Excluded s n outlier y single Grus test.
1082 ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 Tle 4. Comprison of chemicl nlysis nd FoodScn methods Rnges of sttisticl prmeters Method s r RSD r, % s R RSD R, % Protein Chemicl nlysis Study smples using 992.15 0.12 0.60 0.49 3.00 AOAC 992.15 collortive study 0.12 0.41 0.60 2.23 0.18 0.46 1.32 3.35 FoodScn Collortive study dt 0.07 0.33 0.35 2.13 0.11 0.79 0.54 5.23 SEP (from precollortive vlidtion) 0.62 Ft Chemicl nlysis Study smples using 960.39 0.10 1.17 0.84 9.46 AOAC 960.39 collortive study 0.12 2.48 1.71 8.75 0.15 3.13 2.26 11.05 FoodScn Collortive study dt 0.05 0.28 0.22 2.67 0.13 0.44 0.52 6.89 SEP (from precollortive vlidtion) 0.78 Moisture Chemicl nlysis Study smples using 950.46B() 0.22 0.83 0.30 2.87 ISO 1442 collortive study 0.24 0.26 0.34 0.54 0.36 0.41 0.55 0.81 FoodScn Collortive study dt 0.13 0.56 0.23 0.92 0.16 0.72 0.39 1.55 SEP (from precollortive vlidtion) 0.72 Smple 5: Blood leked from dish to plstic g ; Smple 17: Moisture leked from dish to plstic g. These smples were scnned, nd the results were reported with those for the remining study smples. Collortor 11 initilly rn smples, using n incorrect FoodScn profile; smples were immeditely rescnned under the correct profile, nd the results were reported to the Study Director. Collortor 15 rn Smple 9 twice nd reported the results s those for Smples 9 nd 10. Smple 10 ws run, nd the results were sumitted to the Study Director. Both of these smple-hndling errors were immeditely discovered upon initil review of the dt. Recommendtions The Study Director recommends tht this method for the Determintion of Ft, Moisture, nd Protein in Met nd Met Products y Using the FOSS FoodScn Ner-Infrred Spectrophotometer with the FOSS Artificil Neurl Network (ANN) Clirtion Model nd Associted Dtse e dopted s Officil First Action. Acknowledgments The Study Director thnks Michel Busch, Mrk Host, Jeff Boedigheimer, Andree Behrens, nd Christin Hrpsøe (FOSS North Americ, Eden Pririe, MN) for their ssistnce with preprtion of the prctice nd study smples nd Jko Jons (FOSS Anlyticl, Hillerod, Denmrk) for contriuting smple cups for the study. The Study Director lso thnks the following collortors for their prticiption in this study: Susn Aldn nd Ashley Perez, Beef Products Inc., Holcom, KS John Birkey, Crgill Met Solutions, Frion, TX Ryn Conwy, Crgill Met Solutions, Ft. Morgn, CO Tony Hemminger nd Todd Shinkle, Pierre Foods, Cincinnti, OH Jeremy Herrick, Hormel Foods, Algon, IA Crystl Hurley, SSI Foods, Inc., Wilder, ID Jennifer Longind nd Mrshll Beggs, Beef Products Inc., Amrillo, TX Scott McMignl, Beef Products Inc., South Sioux City, NE
ANDERSON: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 4, 2007 1083 Kevin Nnke nd Tony Mitchell, Lopez Foods, Oklhom City, OK Mr Phelps, Cristin Ionit, Andi Milu, Jmes Willims, nd Andres Conquest, Tyson Fresh Mets Inc., Goodlettsville, TN Josh Reitz, Beef Products Inc., Wterloo, IA Andy Slzr, Crgill Met Solutions, Plinview, TX Mtthew Strmpe nd Krist Vn Horn, Hormel Foods, Austin, MN Steve Zelten nd Christy Wipperfurth, Fir Oks Frms, Plesnt Pririe, WI Steve Zerr, Cloverdle, Mndn, ND References (1) Borggrd, C., & Thoderg, H.H. (1992) Anl. Chem. 64, 545 551 (2) Borggrd, C. (2001) in Ner Infrred Technology in the Agriculture nd Food Industries, P. Willims & K. Norris (Eds), Americn Assocition of Cerel Chemists, Inc., St. Pul, MN, pp 101 107 (3) Mrk, H., & Workmn, J. (1991) Sttistics in Spectroscopy, Acdemic Press, Sn Diego, CA, p. 299 (4) Mrk, H., & Workmn, J. (1991) Sttistics in Spectroscopy, Acdemic Press, Sn Diego, CA, p. 302 (5) Willims, P.C. (2001) in Ner Infrred Technology in the Agriculture nd Food Industries, P. Willims & K. Norris (Eds), Americn Assocition of Cerel Chemists, Inc., St. Pul, MN, pp 162 163 (6) Officil Methods of Anlysis (2000) 17th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method 983.18 (7) Officil Methods of Anlysis (2000) 17th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method 992.15 (8) Officil Methods of Anlysis (2000) 17th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method 960.39 (9) Officil Methods of Anlysis (2000) 17th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method 950.46 (10) Thiex, N.J., Mnson, H., Anderson, S., & Persson, J.A. (2002) J. AOAC Int. 85, 309 317 (11) Foster, M., & Gonzles, S. (1992) J. AOAC Int. 75, 288 292 (12) Youden, W.J., & Steiner, E.H. (1975) Sttisticl Mnul of the AOAC, AOAC INTERNATIONAL, Arlington, VA, p. 74 (13) Lynch, J.M. (2001) AOAC Sttisticl Progrm for Blind Replictes, Version 1.14, Ithc, NY (14) Horwitz, W. (1982) Anl. Chem. 54, 67A 76A (15) Youden, W.J., & Steiner, E.H. (1975) Sttisticl Mnul of the AOAC, AOAC INTERNATIONAL, Arlington, VA, pp 26 27 (16) Officil Methods of Anlysis (2000) 17th Ed., AOAC INTERNATIONAL, Githersurg, MD, Method 972.16 (17) FOSS FoodScn User Mnul (2004) FOSS Anlyticl A/S, Hillerod, Denmrk (18) ISO 1442:1997(E), Met nd Met Products Determintion of Moisture Content (Reference Method), 2nd Ed., Interntionl Orgniztion for Stndrdiztion, Genev, Switzerlnd