24th Int. Symp. “Animal Science Days”, Ptuj, Slovenia, Sept. 21st−23rd, 2016. Acta argiculturae Slovenica, Supplement 5, 127–131, Ljubljana 2016 COBISS: 1.08 Agris category code: L01, Q04 RELATIONSHIP BETWEEN CARCASS LEAN MEAT PERCENTAGE AND MRI DATA OBTAINED ON HAMS OF ENTIRE BOARS Žarko RADIŠIĆ 1, Maren BERNAU 2, Ivona DJURKIN KUŠEC 3, Ulrich BAULAIN 4, Armin M. SCHOLZ 5, Goran KUŠEC 6 Relationship between carcass lean meat percentage and MRI data obtained on hams of entire boars 1 Faculty of Agriculture in Osijek, Josip JurajStrossmayer University of Osijek, Kralja Petra Svačića 1d, 31000 Osijek, Croatia, e-mail: zradisic@pfos.hr 2 Livestock Center Oberschleißheim, Ludwig-Maximilians-University Munich, Sankt-Hubertusstrasse 12, 85764 Oberschleißheim, Germany, e-mail: Maren.Bernau@ lvg.vetmed.uni-muenchen.de 3 Same address as 1, e-mail: idurkin@pfos.hr 4 Institute of Farm Animal Genetics, Friedrich-Loeffler-Institut, Hoeltystr.10, 31535 Neustadt, Germany, e-mail: ulrich.baulain@fli.bund.de 5 Same address as 2, e-mail: Armin.Scholz@lvg.vetmed.uni-muenchen.de 6 Same address as 1, e-mail: gkusec@pfos.hr ABSTRACT The study on the relationship between lean meat percentage (LMP) and ham measures assessed by means of mag- netic resonance imaging (MRI) was performed on 20 carcasses of entire boars from one performance testing station in Germany. In addition, linear carcass measurements were taken, according to the standard protocol for stationary performance testing, to be used to estimate LMP by the Bonner Formula. Finally, manual dissection was serving as reference, for LMP. The correlations between the true LMP and linear measurements Bonner Formula were medium strong, but not in all cases statistically significant. On the other hand, the correlations between ham data obtained by MRI and reference LMP had stronger and significant correlations. Three new formulas were constructed in order to evaluate the LMP of entire boars. The best prediction ability was found when combining linear measurements with one MRI measure – the widest circumference of the ham (R2 = 0.67 and RMSEP = 1.395). Interestingly, the proposed equa- tion has variables that can be easily taken; three of them are taken regularly for Bonner Formula estimation and MRI based ham circumference aided to improve the prediction accuracy, a measure that could also be acquired by the simple use of meter or rope. Key words: pigs, entire boars, lean meat percentage, Magnetic resonance imaging, dissection 1 INTRODUCTION The assessment of commercial value of pig car- cass is a permanent issue in meat industry. It has always been expressed as lean meat percentage (LMP) and this opened a whole area of research aiming at the most accu- rate prediction of this trait. The most precise assessment of pig carcass composition is obtained by manual dissec- tion which is time consuming, expensive and results in the loss of carcass. The latter drawback of this method can nowadays be overcome by the use of non-invasive techniques as computed tomography (CT) or magnetic resonance imaging (MRI). The latter was shown to be a precise tool for pig body composition estimation (Baul- ain, 1997; Mitchell et al., 2001; Baulain et al., 2010). Moreover, some results indicated that scanning of the ham alone can be used as a basis for a good prediction of LMP in pig carcasses and this could reduce the cost of scanning (Font i Furnols et al., 2009). However, due to the speed of modern slaughter lines, LMP is most often assessed by the use of on-line methods. Most of those are based on measurements taken on the carcass involving various thicknesses of fat layer and muscle depths, although there were some attempts of us- age of other independent variables; e.g. fat and muscle surfaces measured at various positions of pig’s carcass (Kušec et al., 2014). High correlation of these measure- Acta agriculturae Slovenica, Supplement 5 – 2016128 Ž. RADIŠIĆ et al. ments with the actual LMP is a critical criterion for their employment. Bonner formula, developed in Germany for prog- eny/performance testing of barrows and gilts, is a good example of using highly correlated carcass measurements for LMP prediction (Tholen et al., 2004). However, Bon- ner Formula is still not developed for entire boars. In near future (2019 in EU) the carcasses of entire male pigs will appear more frequent at the slaughter lines which means that an accurate method of their LMP prediction will be needed soon. The aim of this study is to establish the correlations between LMP in the carcasses of entire male pigs and car- cass measurements as well as some MRI data obtained on their hams in order to investigate the possibility of LMP predictions to be used at the slaughterlines. 2 MATERIAL AND METHODS The present study was carried out on 20 carcasses of entire boars all of them being F1 crossbreds from sire line Pietrain and Landrace x Large White dams. The boars were kept in single pens at one performance test station in Germany, fed ad libitum with a station specific diet. After reaching the intended slaughter weight of approxi- mately 110 kg, boars were slaughtered in the abattoir connected to the test station. After slaughter and primary processing of the carcasses, linear measurements on the right carcass side were taken according to the standard protocol (Richtlinie ….., 2007) as follows: – Loin and back fat area – LA and BFA (cm2), re- corded at the cut between 13th and 14th thoracic vertebrae – side fat thickness – SF (cm), measured ventral to the Musculus latissimus dorsi perpendicular to the rind between the 13th and 14th thoracic ver- tebrae – Backfat thickness, specified as an average of three measured fat thicknesses (the fattest at the with- ers – BFW, the thinnest in the middle of the back – BFM, and the one over the lumbar muscles – BFL) – Speckmaβ B – SMB (cm), determined at the chop angle describing the thinnest part of the fat pad between 13th and 14th thoracic vertebrae These measures were used for LMP prediction in the pig carcasses using the Bonner Formula, as the per- formance test protocol prescribes. MR imaging (MRI) of the hams was performed us- ing a Siemens Magnetom Open tomograph which is an open low-field magnetic system with field strength of 0.2 T. Transverse sections were acquired using following pa- rameters: repetition time 700 ms, echo time 17 ms, 7 mm slice thickness, 0.7 mm distance between slices, 18 slices, flip angle of 90°, 458 mm field of view, 256*256 matrix size and 1 accumulation. The sequence was positioned vertical to the femur to cover the surrounding muscles. By the use of MRI following traits of the ham were ob- tained: widest circumference of the ham – CIRC (mm), surface of slice at the widest circumference – SURF (cm2), estimated ham volume – estV (mm3), volume of the muscles in ham – hamM (mm3), volume of the fat in ham – hamF (mm3). Finally, the right carcass sides were submitted to the total dissection into the main tissues: muscle (lean meat), fat, and bone, providing the actual lean meat percentage (LMP) as a reference. Statistical analysis was performed with statistical program Dell Statistica (data analysis software system), version 12. software.dell.com (Dell Inc. 2015); mod- ules Basic Statistics and Tables – Correlation matrices and module Multiple Regression. The full cross valida- tion procedure and graphs were made by CAMO Software AS, NedreVollgate 8, N-0158 OSLO, Norway. 3 RESULTS AND DISCUSSION Linear carcass measures used as pre- dictors in Bonner Formula and their rela- tionship with true LMP obtained by dis- section are presented in Table 1. It is obvious that backfat measures, BFW and BFM, are not significantly cor- related with true LMP. This is not making them the best choice as predictors of LMP in the carcasses of entire male pigs used Variable Mean Standard Deviation Correlation to LMP Lean meat percentage, LMP (%) 60.94 2.44 - Back fat area, BFA (cm2) 12.55 2.31 −0.53* Loin area, LA (cm2) 51.65 4.75 0.66* Side fat, SF (cm) 2.11 0.33 −0.57* Back fat whiters, BFW (cm) 2.87 0.35 −0.33 Back fat middle of the back, BFM (cm) 1.51 0.20 −0.30 Back fat lumbar, BFL (cm) 0.75 0.23 −0.52* Speckmass B, SMB (cm) 0.83 0.20 −0.63* Table 1: Descriptive statistics for measures used in Bonner Formula for carcass assessment and their correlation with LMP * marked correlations are statistically significant (p < 0.05) Acta agriculturae Slovenica, Supplement 5 – 2016 129 RELATIONSHIP BETWEEN CARCASS LEAN MEAT PERCENTAGE AND MRI DATA OBTAINED ON HAMS OF ENTIRE BOARS in this study. Just for illustration, on the basis of 20 boar carcasses investigated in this study, a new formula for the prediction of LMP was derived: LMP = 52.51639 − 0.377 * BFA + 0.327 * LA − 3.825 * BFL − 1.721 * BFM + 1.309 * BFW − 1.519 * SF + 1.437 * SMB The regression of LMPs predicted by this equation on actual LMP obtained by the dissection of entire boar carcasses is shown in Figure 1. Parameters indicating the accuracy of LMP prediction had favourable values, par- ticularly coefficient of determination (R2) and root mean square error (RMSEP) which were in present case 0.49 and 1.70, respectively. On the other hand, data obtained by MR imaging of the hams have shown stronger correlations, all being significant, with the LMP obtained by dissection of the carcasses used in this study (Table 2). Such strong correlations make these measurements good predictors of LMP in the carcasses of investigated entire boars. Regression equation constructed for LMP prediction in this case was: LMP = 61.60681 − 0.04174 * CIRC + 0.00091 * SURF −0.00151 * estV + 0.00323 * hamM − 0.01499 * hamF The prediction of the LMP in the carcasses of investi- gated entire boars by this formula is presented on Figure 2. Prediction capacity of the formula constructed by the use of MRI measurements as independent variables was improved in respect to a new formula derived from the variables used in the Bonner Formula. This is appar- ent from the increased values of R2 (0.67) which explains the total variation of the actual LMP and lowered RMSEP (1.359) indicating an error of the prediction. A search for the different new measures that could improve the accuracy of LMP prediction are of permanent interest of researchers. The variety of investigating ap- proaches in this matter is wide, from measuring surfaces of fat and muscle at specific places by MRI like Kremer et al. (2013) to pure drawing on transparent paper as per- formed by Kušec et al. (2014). MRI technique is usually not feasible for the routine as- sessment of LMP, but measure- ment such as widest circumfer- ence of the ham, which showed fairly good correlation to LMP as presented here, could also be taken without the use of such expensive method. Such vari- able could be used in order to improve the accuracy of a more practical LMP formula, instead Figure 1: LMP prediction in the carcasses of investigated entire boars by a new formula based on all variables used in the Bonner Formula Variable Mean Standard Deviation Correlation to LMP Lean meat percentage, LMP (%) 60.94 2.44 – Widest circumference, CIRC (mm) 850.89 27.96 0.60* Surface at the widest circumference, SURF (mm2) 43535.77 2946.73 0.70* Estimated ham volume, estV (mm3) 5317.93 402.67 0.63* Muscle volume in ham, hamM (mm3) 3410.18 258.37 0.82* Fat volume in ham, hamF (mm3) 514.02 79.04 −0.53* Table 2: Descriptive statistics for traits of ham measured by MRI and their correlation with LMP * marked correlations are statistically significant (p < 0.05) Acta agriculturae Slovenica, Supplement 5 – 2016130 Ž. RADIŠIĆ et al. of the variables that showed insignificant correlations with actual LMP obtained by dissection of the boar carcasses. In present study we also attempted to improve the ability of LMP prediction by equations using various combinations of independent variables from Bonner Formula and MRI measurements of the ham. Figure 3 shows the equation that yielded the best in- dicators of accuracy (R2 = 0.67 and RMSEP = 1.395). For this equation, only linear measurements taken directly (BFA, LA, BFL) or virtually (CIRC) from the carcass were used, resulting in the following parameters: LMP = 32.34091 − 0.28296 * BFA + 0.23627 * LA − 3.95181 * BFL + 0.02690 * CIRC Of all measurements obtained by the use of MRI, only ham circumference (CIRC) aided in the improve- ment of predicting ability, but this feature of the ham can be also measured by the simple use of meter or rope. Attempts to improve the Bonner formula to assess carcass lean meat content in entire boars by use of non- invasive methods were made before. Recently, Bernau et al. (2015) investigated the use of dual energy X-ray absorptiometry (DXA), magnetic resonance imaging (MRI) and dissection to evaluate the carcass composi- tion; and they also tested the suitability of the Bonner Formula. The authors found that formulas from DXA, MRI and combination of MRI/DXA yielded the same and better accuracies than Bonner Formula. However, their result was obtained on the same sample of entire boars’ carcasses (n = 20) as in the present study, meaning that these results should be interesting enough to be repeated Figure 2: LMP prediction in the carcasses of entire boars by the equation constructed by use of MRI measurements as predicting variables Figure 3: LMP prediction in the carcasses of investigated entire boars by the adapted LMP formula with MRI measurement of ham circumference as additional predicting variable Acta agriculturae Slovenica, Supplement 5 – 2016 131 RELATIONSHIP BETWEEN CARCASS LEAN MEAT PERCENTAGE AND MRI DATA OBTAINED ON HAMS OF ENTIRE BOARS on a larger sample. This can also be stated for the results of the investigations presented here. 4 CONCLUSION The results from present study show that the meas- urements such as BFW and BFM were not significantly correlated with the actual LMP, which reduces the ac- ceptability of Bonner Formula as method for prediction of LPM in the carcasses of entire boars. On the other hand, the measures of the ham as- sessed by MR imaging had stronger and significant cor- relations with LMP obtained by dissection. The formula constructed using MRI measurements as independent variables had increased prediction ability in respect to the “new” Bonner Formula. Equation taking into account BFA, LA and BFL measures together with ham circumference obtained by MRI (CIRC) as independent variables, had the best pre- dicting ability. Since ham circumference can be also measured us- ing meter or rope it could be considered as an interesting predictor for future studies of methods for the prediction of LMP in pig carcasses from entire boars. 5 REFERENCES Baulain, U., (1997). Magnetic resonance imaging for the in vivo determination of body composition in animal science. Computers and Electronics in Agriculture, 17, 189–203. Baulain, U., Friedrichs, M., Hoereth, R., Henning, M., Tholen, E. (2010). Use of MRI to assess carcass and primal cut com- position in different pig breeds. Retrieved from http:/www. kongressband.de/wcgalp2010/assets/pdf/0357.pdf. Bernau, M., Kremer, P.V., Lauterbach, E., Tholen, E., Petersen, B., Pappenberger, E., Scholz, A.M. (2015). Evaluation of carcass composition of intact boars using linear measure- ments from performance testing, dissection, dual energy X-ray absorptiometry (DXA) and magnetic resonance im- aging (MRI). Meat Science 104, 58–66. Dell Inc. (2015). Dell Statistica (data analysis software system), version 12. software.dell.com. Font i Furnols, M., Teran, M.F., Gispert, M. (2009). Estimation of lean meat content in pig carcasses using X-ray Computed Tomography and PLS regression. Chemometics and Intel- ligent Laboratory Systems, 98, 31–37. Kremer, P.V., Foerster, M., Scholz, A. M. (2013). Use of mag- netic resonance imaging to predict the body composition of pigs in vivo. Animal, 7(6), 879–884. Kušec, G., Djurkin Kušec, I., Radišić, Ž. (2014). Backfat and muscle surface measurements as predictors of lean meat percentage in pig carcasses. In C. Maltin, C. Craigie & L. Bunger (Eds.), Farm Animal Imaging Copenhagen 2014 (pp. 28–30). Scotland’s Rural College. Mitchell, A.D., Scholz, A.M., Wang, P.C., Song, H. (2001). Body composition analysis of the pig by magnetic resonance im- aging. Journal of Animal Science, 79, 1800–1813. Richtlinie für die Stationsprüfung auf Mastleistung, Schlachkörperwert und Fleischbeschaffenheit beim Sch- wein (2007) .. www.zds-bonn.de/services/files/dokumente/ rlnkp07.pdf Tholen, E., Wiese, M., Baulain, U., Hoereth, R., Hoppenbrock, K.H. (2004). Schätzung des Muskelfleischanteils – Unter- suchung von stationär leistungsgeprüften Zuchtschweine- herkünften. Fleischwirtschaft, 2, 105–110.