Ann. Anim. Sci., Vol. 11, No. 3 (2011) 357 369 Possibilities of using ultrasonography in breeding work with pigs. Part III Estimation of carcass meat content using regression equations developed from ultrasonographic measurements M i r o s ł a w T y r a, M a g d a l e n a S z y n d l e r - N ę d z a, R o b e r t E c k e r t Department of Animal Genetics and Breeding, National Research Institute of Animal Production, 32-083 Balice n. Kraków, Poland Abstract The aim of the study was to develop regression equations for estimating carcass meat percentage and to evaluate their practical usefulness. In addition, relationships between meatiness estimated using different methods (measurement techniques) and actual meatiness obtained from detailed dissection were analysed. Subjects were 476 gilts of the Polish Large White (n = 151), Polish Landrace (n = 149), Pietrain (n = 84) and Duroc (n = 92) breeds tested in Polish Pig Performance Testing Stations (SKURTCh) and slaughtered at 100 kg body weight. The analysed techniques were ultrasound measurements used in live testing (Piglog 105) and ultrasonographic measurements (Aloka SSD 500 with a linear probe) taken on live animals and postmortem. Dissection measurements (taken with a caliper and planimeter) and detailed dissection results for carcass meat and fat weight and percentage served as a reference point for these groups of measurements. The highest accuracy of the equations with small errors (below 2.2% for postmortem ultrasonographic measurements and below 2.7% for live ultrasonographic measurements) was obtained when the equation included the following ultrasonographic measurements: backfat thickness at P4, fat area, and loin eye height. Comparative analysis of the four different methods for meatiness estimation in relation to actual meatiness evaluated from detailed dissection showed the following accuracy ranking of the equations: station test equation (r P = 0.803); equation based on postmortem Aloka ultrasonographic measurements (r P = 0.788); equation based on live Aloka ultrasonographic measurements (r P = 0.762); and equation based on live Piglog 105 measurements (r P = 0.758). Key words: pig, ultrasonographic measurements, carcass meat content, ALOKA, PIGLOG Breeding progress resulting from selection of the pedigree population of pigs is influenced by the accuracy with which animals are estimated. For this reason, research on improving the estimation methods is oriented towards this direction, among other things. In Poland, the pedigree population is estimated based on live and postmortem evaluation (station testing in Polish Pig Performance Testing Stations, SKURTCh). Live evaluation covers the entire pig population and slaughter
358 M. Tyra et al. value is estimated based on Piglog 105 ultrasonic measurements and a regression equation coded in the device. Apparently, no considerable improvements in the accuracy of this measurement technique should be expected. Progress can be obtained using new instruments that have been used in human and veterinary medicine, namely ultrasonographic devices. The operation of ultrasonographic devices is based on ultrasonic waves. These waves are sent by the transducer, penetrate the body, reflect off at the boundary of different tissues and fluids, and return to the transducer. The number of reflected and captured waves varies according to the acoustic impedance of tissues and fluids. The differences in the acoustic impedance of tissues are used for graphic visualization on the display screen of the measuring device, because reflected waves are picked up by the transducer and converted into an electric signal, which is then amplified and further processed into graphical form (Grygar and Kudlac, 2004). The recent progress that has been achieved in ultrasonography after elaboration of B-mode devices with dynamic imaging (which allows changes in the analysed tissues or organs to be displayed in real time) and the relatively low cost and ease of portability of the equipment made this technique popular with breeders (Tait et al., 2005). However, high-quality images were not recorded and losslessly fed into the computer until the advent of digital video, digital photography and computer-assisted digital image processing (Moeller, 2002). The possibility of digital archiving of the images obtained and their subsequent processing using specialist software (linear dimensioning, calculation of the areas marked on these images) enable some extra parameters to be obtained in addition to backfat thickness and loin eye height. This allows for more objective estimation of fatness and muscling through a greater number of measurement points and higher complexity of measurements, as well as increasing the precision of digital measurements. It is believed that the results obtained with this method will be more accurate than those obtained based on ultrasonic measurements of backfat and muscle thickness or on regression equations, in which half-carcasses have to be partly dissected. The benefit of increased accuracy of measurement should translate into higher relationships between these measurements and actual dissection measurements. This is confirmed by data reported by Forrest et al. (1989) and Moeller et al. (1998), who obtained high relationships between loin eye area measured ultrasonographically and actual area obtained during dissection (r P = 0.58 and r P = 0.68, respectively). The progress achieved with the use of the latest technique is illustrated by our own study (Tyra et al., 2010 b), in which the use of digital dimensioning resulted in higher relationships between ultrasonographic measurements and their dissection equivalents (from r P = 0.625 to r P = 0.793). The objective of the study is to determine the usefulness of the new measurement technique (ultrasonography) for estimation of meatiness and fatness in pigs. The positive results will enable this technique to be used in breeding work with pigs and introduced in station and live testing of animals. The practical usefulness will be ultimately determined by the regression equations developed for meatiness estimation, their accuracy, as well as a comparison of different methods of meatiness estimation (live, ultrasonographic and station testing) with the actual carcass meat content obtained during detailed dissection, which is also the aim of the present study.
Using ultrasonography in breeding work with pigs. Part III 359 Material and methods Animals and data obtained Subjects were gilts of the most common breeds in the Polish pig population. A total of 476 animals representing Polish Large White (n = 151), Polish Landrace (n = 149), Pietrain (n = 84) and Duroc (n = 92) breeds were investigated. At 100 kg body weight, animals were measured for backfat thickness and muscle height using ultrasonic (Piglog 105) and ultrasonographic devices (Aloka SSD 500 with model UST-5011U linear transducer). Animals were then slaughtered and their half-carcasses were chilled at 4 C for 24 h. On the next day, the right half-carcasses were subjected to ultrasonographic measurements (Aloka) and linear dissection measurements. The measurement points were chosen such that as many analogous measurements as possible were collected for each method used (ultrasonic, ultrasonographic and dissection). The following data sets were obtained: ultrasound measurements with a PIGLOG 105 device: P2 backfat thickness behind the last rib, 3 cm off the mid-line, P4 backfat thickness behind the last rib, 8 cm off the mid-line, P4M loin eye height at P4, ultrasonographic measurements with an ALOKA SSD 500 device and model UST- 5011U linear transducer: U2P backfat thickness at P2, U4P backfat thickness at P4, UWP height of m. longissimus dorsi at P4, USP width of m. longissimus dorsi, UOP cross-sectional area of m. longissimus dorsi, UTP cross-sectional area of fat over loin eye area of m. longissimus dorsi,
360 M. Tyra et al. ultrasonographic postmortem measurements with an ALOKA SSD 500 device, taken analogously to live measurement sites: U2D backfat thickness at P2, U4D backfat thickness at P4, UWD height of m. longissimus dorsi at P4, USD width of m. longissimus dorsi, UOD cross-sectional area of m. longissimus dorsi, UTD cross-sectional area of fat over loin eye area of m. longissimus dorsi, dissection measurements: D2 backfat thickness at D2 (equivalent to measurements of P2, U2P and U2D), D4 backfat thickness at D4 (equivalent to measurements of P4, U4P and U4D), DW height of m. longissimus dorsi (equivalent to measurements of P4M, UWP and UWD), DS width of m. longissimus dorsi (equivalent to measurements of USP and USD), DPO planimetered loin eye area (equivalent to measurements of UOP and UOD). The procedures for management of animals and measurements were detailed in an earlier study (Tyra et al., 2010 a). Statistical analysis The correlations estimated between measurements of backfat and longissimus muscle thickness, taken using different measurement methods, and analogous dissection measurements and carcass meat and fat weight and percentage (Tyra et al., 2010 b) were used to choose appropriate traits (parameters) for developing regression equations estimating the meatiness of pigs, which in this case were based on ultrasonographic measurements. A backward stepwise regression was used during construction of the regression equations. It involved constructing a model containing all potential variables followed by their gradual elimination in order to obtain a model with the highest value of the multiple correlation coefficient while preserving the significance of the parameters. The fit of the models to actual data was examined using the multiple correlation coefficient R, which equals the positive root of the coefficient of determination (R 2 ) and is a measure of the strength of the linear relationship between response variable Y and the set of explanatory variables included in the model. R = 1 φ 2 = R 2 R 2 is expressed by the formula: n (ŷ t ȳ) 2 R 2 t=1 = n (y t ȳ) 2 t=1 where: y t actual value of the Y variable, ŷ t theoretical value of the response variable (based on the model), ȳ mean arithmetic value of the response variable.
Using ultrasonography in breeding work with pigs. Part III 361 Then, the equations characterized by the best parameters of R coefficients and the smallest errors were chosen for further analysis from among all the equations developed. These served as a basis for estimating meat percentage for the analysed animals and the equations currently used for live and station testing. The formulas (regression equations) for determining meat percentage in station and live testing are shown below: Regression equations used for estimating carcass meat percentage: actual meatiness estimated based on detailed dissection: %MD = (weight of meat of primal cuts / weight of cold right half-carcass)*100 meat percentage estimated according to the live testing method: %MP = 0.4203 P2 0.4461 P4 + 0.2469 P4M +54.8763 meat percentage estimated according to the method used in Pig Testing Stations: %MS = (((1.745 * ham without backfat and skin) + (0.836* (loin + tenderloin))+ + (0.157*(2 * DS + DW)) 1.884) / weight of cold right half-carcass)*100 where: %MD actual carcass meat content based on detailed dissection of primal cuts, %MP live method for determination of carcass meat content, %MS method used in Pig Testing Stations for determining meat content based on regression equation that includes measurements obtained during simplified dissection. The measurements obtained were used to estimate basic statistical parameters for the meat content results estimated with different measurement techniques. Relationships (correlations) were also determined between meat content estimated using the different measurement techniques and the actual carcass meat content obtained from detailed dissection of the primal cuts. Statistical analysis was performed using the GLM and COR procedure of the SAS statistical package (1989). The statistical model used in the calculations was as follows: Y ijkl = µ + d i + f j + α(x ijk ) + e ijkl where: y ijkl ijkl th observation, µ overall mean, d i effect of i th breed, f j effect of j th sire, α(x ijk ) covariance on right half-carcass weight, e ijkl random error.
362 M. Tyra et al. Differences between the means for individual breeds were tested at the level of 5% and 1% using Duncan s multiple range test. Results Table 1 contains data on the relationships between live and postmortem ultrasonographic measurements taken with an Aloka 500 device and actual carcass meat percentage determined based on detailed dissection. All the estimated coefficients of correlation were significant (P 0.01). The highest relationship with carcass meat percentage was found for measurements of backfat thickness at U4P (r P = 0.711) and U4D (r P = 0.714) and for loin eye measurements UOP (r P = 0.633) and UOD (r P = 0.651). The lowest relationships with this trait were found for measurements of the width of m. longissimus dorsi USP and USD (r P = 0.407 and r P = 0.473, respectively). Table 1. Correlations between ultrasonographic measurements with Aloka SSD 500 and actual carcass meat content determined based on detailed dissection Ultrasonographic measurements with ALOKA 500: %MD U2P U4P UWP USP UOP UTP live 0.531 0.711 0.534 0.407 0.633 0.518 postmortem U2D U4D UWD USD UOD UTD 0.557 0.714 0.569 0.473 0.651 0.548 %MD actual carcass meat content based on detailed dissection, P4M loin eye height at P4, UWP height of m. longissimus dorsi at P4, USP width of m. longissimus dorsi, U2P backfat thickness at P2, U4P backfat thickness at P4, UOP cross-sectional area of m. longissimus dorsi, UTP cross-sectional area of fat over loin eye area of m. longissimus dorsi, UWD height of m. longissimus dorsi, USD width of m. longissimus dorsi, U2D backfat thickness at P2, U4D backfat thickness, equivalent to live backfat thickness measurement point P4, UOD cross-sectional area of m. longissimus dorsi, UTD cross-sectional area of fat over loin eye area of m. longissimus dorsi. Table 2 presents the equations developed separately for live and postslaughter measurements. When deriving the multiple regression equation to estimate carcass meat percentage, we chose the same number of independent variables as in the existing equation used in live animal testing. For comparison, we also presented equations using all the data available. Coefficients of multiple correlation were also given for regression equations estimating carcass meat percentage (R) and equation estimation errors (SE). The coefficients of multiple correlation for all the equations included in this table were significant (P 0.01). The first equations presented for live and postmortem measurement technique accounted for all the measurement points (obtained using ultrasonography). The coefficient of multiple correlation for these equations was the lowest of all those presented (R = 0.76917 for %MUD 1 equation
Using ultrasonography in breeding work with pigs. Part III 363 and R = 0.75187 for %MUP 1 equation). In addition, this equation was associated with the largest error of SE = 3.28 and SE = 3.52, respectively. In the regression equations derived on the basis of analogous measurement points of backfat thickness and muscle that are currently used in live animal evaluation (P2, P4, P4M), the coefficient of multiple correlation R was 0.77477 for ultrasonographic measurements of the half-carcass (U2D, U4D, UWD) and 0.76417 for measurements on live animals (U2P, U4P, UWP), with the standard error of 2.66 and 3.29, respectively. Table 2. Proposed regression equations for estimating carcass meat percentage based on ultrasonographic measurements (ALOKA SSD 500) Regression equations R SE including postmortem measurements % MUD 1 = -.7483 U2D -.6902 U4D +.3201 UWD +.0439 USD -.2391 UTD +.1444 UOD + 54.26 0.76917 3.28 % MUD 2 = -.4892 U2D - 2.464 U4D + 2.5264 UWD +49.92 0.77477 2.66 % MUD 3 = - 2.707 U4D -.2536 UTD +.1892 UOD + 54.67 0.81307 2.22 including live measurements % MUD 4 = -.8107 U4D -.1436 UTD +.3689 UWD + 56.42 0.81517 2.13 % MUP 1 = -.6813 U2P -.9512 U4P +.2931 UWP +.04112 USP -.2177 UTP +.2331 UOP + 53.76 0.75187 3.52 % MUP 2 = -.4972 U2P - 2.6431 U4P + 2.2364 UWP +49.27 0.76417 3.29 % MUP 3 = - 2.5847 U4P -.2481 UTP +.1767 UOP + 54.25 0.79587 2.85 % MUP 4 = -.8623 U4P -.1517 UTP +.3193 UWP + 56.76 0.80847 2.63 Meat percentage estimated with different methods: %MUP (1, 2, 3, 4) ultrasonographic method based on dimensioning of images obtained on live animals, %MUD (1,2,3,4) ultrasonographic method based on results obtained from images recorded postmortem, Measurements: U2P backfat thickness at P2, U4P backfat thickness at P4, UOP cross-sectional area of m. longissimus dorsi, UTP cross-sectional area of fat over loin eye area of m. longissimus dorsi, UWD height of m. longissimus dorsi, USD width of m. longissimus dorsi, U2D backfat thickness at P2, U4D backfat thickness, equivalent to live backfat thickness measurement point P4, UOD cross-sectional area of m. longissimus dorsi, UTD cross-sectional area of fat over loin eye area of m. longissimus dorsi. Using multiple regression and the stepwise elimination option, the following variables for estimating carcass meat percentage were chosen: backfat thickness at U4 (U4D and U4P), fat area (UTD and UTP) and height of loin eye (UWD, UWP). The regression equations that included postmortem (%MUD 4 ) and live measurements (%MUP 4 ) were characterized by the highest multiple correlation of R = 0.81517 for %MUD 4 and R = 0.80847 for %MUP 4 and the smallest error of 2.13% and 2.63%, respectively (equations marked with grey shading).
364 M. Tyra et al. Table 3. Basic statistical characteristics for meat percentage estimated with different measurement techniques in different pig breeds Measurements Breeds x δ min max v %MD PLW 58.6 2.74 45.5 65.4 4.68 PL 58.8 2.43 45.6 63.3 4.13 Pietrain 66.9 2.14 54.3 71.2 3.20 Duroc 59.6 2.82 46.0 62.3 4.87 TOTAL 60.6 3.13 45.5 71.2 5.17 %MP PLW 55.8 3.28 43.8 61.4 5.88 PL 56.1 2.98 53.0 62.6 5.31 Pietrain 62.3 2.59 54.2 68.9 4.16 Duroc 57.3 2.12 51.3 59.8 3.90 TOTAL 57.1 3.94 43.8 68.9 6.91 %MUP PLW 56.8 2.28 45.9 64.9 4.01 PL 56.9 3.18 46.7 63.3 5.59 Pietrain 63.3 3.59 53.2 69.9 5.67 Duroc 57.3 3.12 48.0 61.6 5.75 TOTAL 57.7 3.94 45.9 69.9 6.83 %MUD PLW 57.5 2.56 45.9 64.9 4.45 PL 57.8 2.76 46.7 63.3 4.78 Pietrain 64.2 3.21 53.2 69.9 5.00 Duroc 58.7 3.36 48.0 61.6 6.14 TOTAL 58.5 3.52 45.9 69.9 6.02 %MS PLW 57.8 2.81 47.2 67.6 4.86 PL 58.2 3.54 49.3 68.5 6.08 Pietrain 65.3 2.98 53.6 73.4 4.56 Duroc 59.1 3.14 47.7 63.5 5.60 TOTAL 59.2 3.58 47.2 73.4 6.05 Meat percentage estimated with the following methods: %MD actual carcass meat content estimated based on detailed dissection, %MP live method, %MUP ultrasonographic method based on dimensioning of images obtained on live animals, %MUD ultrasonographic method based on results obtained from images recorded postmortem, %MS method used in Pig Testing Stations (using regression equation that includes simple dissection measurements). Table 3 shows the basic statistical parameters for carcass meat percentage estimated using the dissection (%MD), live (%MP) and ultrasonographic methods based on the results of dimensioning images obtained from live animals (%MUP), using the ultrasonographic method based on the results obtained from images recorded postmortem (%MUD) and using the method applied in Pig Testing Stations (%MS). The results of this analysis indicate that carcass meat percentage was slightly understated by the live method (by about 3.5% on average) and the ultrasonographic meth-
Using ultrasonography in breeding work with pigs. Part III 365 od involving live animals (by about 2.9%). The equation used in Pig Testing Stations gave 1.3% lower meatiness on average compared to actual measurements. Similar mean meatiness to the station testing method was obtained based on the equation that included ultrasonographic measurements taken on half-carcasses (%MUD). The coefficients of variation for these two measurement methods fluctuated around 6%. For the equations discussed above (%MUP and %MP), the mean variation of about 7% was observed. The differences in meat percentage between the presented methods were not significant. Table 4. Coefficients of correlation between carcass meatiness estimated by different methods and meatiness determined based on detailed dissection Breeds %MP %MUP %MUD %MS %MD PLW.778.815.826.835 PL.793.783.822.816 Pietrain.682.667.693.743 Duroc.742.756.752.758 Total.758.762.788.803 Meat percentage estimated with the following methods: %MD actual meatiness estimated based on detailed dissection, %MP live method, %MUP ultrasonographic method based on dimensioning of images obtained on live animals, %MUD ultrasonographic method based on results obtained from images recorded postmortem, %MS method used in Pig Testing Stations (using regression equation that includes simple dissection measurements). Table 4 presents the coefficients of correlation between meatiness estimated based on chosen equations and meatiness determined based on dissection. The analysis of these relationships indicates that the equation used in Pig Testing Stations (%MS) comes closest to the actual meatiness of the analysed pigs determined based on detailed dissection (r P = 0.803). The results obtained ultrasonographically using postmortem images are close to those obtained using the station method (r P = 0.788) and slightly more accurate than the results of live testing (r P = 0.758) and ultrasonographic testing of live animals (r P = 0.762). Discussion The estimated simple correlations were used to choose traits (measurement points) included in the estimation of carcass meat percentage using a regression equation. The estimation of meatiness based on several measurements makes the evaluation more accurate. For this reason, in our study we estimated the coefficients of multiple regression using ultrasonographic measurements highly correlated with carcass muscling. The presented regression equations for estimation of carcass meat percentage based on live and postmortem ultrasonographic measurements are characterized by high accuracy. The coefficient of multiple correlation for these equations exceeded R = 0.75 and the error of equations with three variables ranged from 2.22 to 3.29. Similar accuracy of estimating carcass meat percentage using an
366 M. Tyra et al. Aloka device was reported by Dioon et al. (1996) for measurements obtained along m. longissimus dorsi (R = 0.8062). For the measurement technique used in the present study, i.e. transversely across m. longissimus dorsi, this coefficient was reported by the authors cited above to be slightly lower (R = 0.741). A much lower coefficient of multiple correlation (R = 0.6793) for meatiness estimation was obtained by Arkide et al. (1992), who used an older type of measuring device (Aloka 210-DX) with a 4-bit resolution. This example illustrates how the advances made in this measuring technique increase the accuracy of measurement. The application of the latest devices that use magnetic resonance allows carcass meat weight to be estimated with an accuracy of R = 0.90 (Berg et al., 1998), but this expensive technique is only used experimentally and probably will never be used in breeding practice on a wide scale. In theory, after increasing the number of variables the regression equation should estimate traits with greater accuracy. This is confirmed by the findings of Terry and Savell (1998). Of the eleven equations proposed by the authors for estimating carcass meat percentage in primal cuts, the coefficients of multiple correlation (R) for these equations increased from 0.82 to 0.95 with the increase in the number of variables included in these equations. It should be noted that the same authors obtained R = 0.9 for the equation accounting for as many as eight variables, which considerably limits the practical application of such an equation. This is not supported by our results, where the increase in the number of variables did not make these equations more accurate, but this was probably due to the low correlations between individual measurement points within ultrasonographic method (live and postmortem). Likewise, Schinckel et al. (2000) and Hick et al. (1998) reported high coefficients of multiple correlation (from R = 0.825 to R = 0.930). These coefficients were calculated for equations estimating carcass meat content in gilts and barrows, but standard errors for the equations derived by Schinckel et al. (2000) were high at SE = 6.86 for equations with two variables and SE = 4.32 for equations with three variables, which is also in agreement with the rule formulated above. The accuracy of the equations can be increased not only by the introduction of more variables. Liu and Stouffer (1995) obtained high values of multiple correlations for equations proposed to estimate carcass meat percentage in primal cuts of the half-carcass (R = 0.883) and carcass meat percentage in the loin (R = 0.803). It should be pointed out that these equations accounted for only two variables: backfat thickness and height of m. longissimus dorsi measured with an Aloka 500V device. Probably, this accuracy of the equations was obtained by averaging the values of the above characteristics based on a series of five consecutive measurements taken for each measurement point. Studies in this area are not often published because positive results that may be used in practice are usually patented and only known to patent applicants or producers of the measuring instrument, whereas publicity or implementation material provide no such data. There are also no studies to compare different measurement techniques in this area. The producers of these instruments are usually not interested in making such comparisons. The present study also attempted to compare four methods of estimating meatiness in pigs: the method used in live testing, the method used in Pig Testing Stations, and the equation that includes ultrasonographic measurements in relation to the results obtained based on detailed dissection. The estimated relationships provided
Using ultrasonography in breeding work with pigs. Part III 367 information on the usefulness of different regression equations (and thus different measurement techniques) for estimating carcass meat percentage while enabling the accuracy of each technique to be compared separately. During the estimation of carcass meat percentage, greater accuracy was obtained when live animals were examined using an Aloka ultrasonographic device (%MUP) compared to the live testing method in current use (%MP). More detailed analysis of these results with regard to breeds revealed the lowest correlations in this regard for Pietrain animals. These disproportions between the Pietrain and other breeds were particularly evident for meat percentage estimated with the equation used in live testing (%MP) and the equation developed for live ultrasonographic measurements (%MUP), but the decrease in this relationship was in each case influenced by different factors. In the live testing, it resulted from the estimations by the Piglog 105 device (and the programmed regression equation) being understated for high-lean animals (Blicharski and Ostrowski, 1996), whereas for ultrasonographic measurements this was due to the construction of the measuring probe (Tyra et al., 2010 a) and can be fairly easily eliminated (Tyra et al., 2010 b). This means that ultrasonographic devices can soon replace live testing devices (Piglog 105). The obstacles are high price and lower portability of the latter instruments, as well as the need to use a computer and specialist software to dimension some characteristics (loin eye area and fat area). It is anticipated that this measurement technique will be used in testing stations. This would enable half-carcass dissection (which is labour-consuming, expensive and disqualifies half-carcasses as full-value commodity) to be eliminated or limited to several measurements a tendency observed in many countries in which station testing is used. This method would have to be fine-tuned to increase the accuracy of meatiness estimation by analysing a greater number of measurement points. It is believed that positive results that increase equation accuracy can be achieved by introducing one or two parameters obtained during dissection measurements, as shown by Liu and Stouffer (1995). Cisneros et al. (1996) obtained a coefficient of multiple correlation of R = 0.9237 for estimating carcass meat percentage after introducing weight at slaughter and two ultrasonographic measurements (height and cross-sectional area of m. longissimus dorsi) into the equation. It should be noted that the objective of this study was to analyse the usefulness of ultrasonography in breeding practice. For this reason, the equations developed and compared only included measurements obtained from this measuring device (Aloka). The positive results of this analysis make it possible to search for methods that increase the accuracy of these equations by the use of measurement points obtained from different sources (multiple techniques): ultrasonography, dissection and live measurements, which could be the subject of further studies in this area. The high values of multiple correlations, the low errors of the equations developed on the basis of ultrasonographic measurements, and the high relationships between meatiness estimated based on these equations and the actual carcass meatiness determined by dissection lead us to conclude that the ultrasonographic method could be used in breeding practice, in particular for live estimation. Moreover, if efforts are made to improve the technical aspects of the measuring probe and to popularize such instruments by reducing their price, in the near future they can supplant the
368 M. Tyra et al. ultrasonic devices currently used for live estimation (Piglog 105). In the case of station testing, the advantage of the ultrasonographic method over the station method was not observed. However, the introduction of these devices into station testing would help to reduce animal dissection costs without compromising the accuracy of estimation. References A r k i d e J.T., B r o r s e n B.W., W h i p k e r L.D., F o r r e s t J.C., K u e i C.H., S c h n i c k e l A.P. (1992). Evaluation of alternative techniques to determine pork carcass value. J. Anim. Sci., 70: 18 28. B e r g E.P., E n g e l B.A., F o r r e s t J.C. (1998). Pork carcass composition derived from a neural network model of electromagnetic scans. J. Anim. Sci., 76: 18 22. B l i c h a r s k i T., O s t r o w s k i A. (1996). Usefulness of ultrasound equipment Piglog 105 to assess the quality of pig carcass (in Polish). Prac. Mat. Zoot., 48: 23 29. C i s n e r o s F., E l l i s M., M i l l e r K.D., N o v a k o f s k i J., W i l s o n E.R., M c K e i t h F.K. (1996). Comparison of transverse and longitudinal real-time ultrasound scans for prediction of lean cut yields and fat-free lean content in live pigs. J. Anim. Sci., 74: 2566 2576. D i o o n N., P e t t i g r e f D., D a i g l e J.P. (1996). Efficiency of various live measurements for the prediction of lean yield and marbling. Centre de Developpement du Porc du Quebec inc. (CDPQ). (http://www.nsif.com/conferences/1996/dion.html). F o r r e s t J.C., K u e i M.W., O r c u t t A.P., S c h n i c k e l J.R., S t o u f f e r J.R., J u d g e M.D. (1989). A review of potential new methods of on-line pork carcass evaluation. J. Anim. Sci., 67: 2164 2170. G r y g a r I., K u d l a c E. (2004). Ultrasound in obsterics and veterinary gynecology (in Polish). Wyd. Platan, 251 pp. H i c k C., S c h i n c k e l A.P., F o r r e s t J.C., A k r i d g e J.T., W a g n e r J.R., C h e n W. (1998). Biases associated with genotype and sex in prediction of fat-free lean mass and carcass value in hogs. J. Anim. Sci., 76 : 2221 2234. L i u Y., S t o u f f e r J.R. (1995). Pork carcass evaluation with an automated and computerized ultrasonic system. J. Anim. Sci., 73: 29 38. M o e l l e r S.J., C h r i s t i a n L.L., G o o d w i n R.N. (1998). Development of adjustment factors for backfat and loin muscle area from serial real-time ultrasonic measurements on purebred lines of swine. J. Anim. Sci., 76: 2008 2016. M o e l l e r S.J. (2002). Evolution and use of ultrasonic technology in the swine industry. J. Anim. Sci., 80 (E. Suppl. 2): E19 E27. S c h i n c k e l A.P., W a g n e r J.R., F o r r e s t J.C., E i n s t e i n M.E. (2000). Evaluation of alternative measures of pork carcass composition. J. Anim. Sci., 79: 1093 1119. T a i t R.G. Jr, W i l s o n D.E., R o u s e G.H. (2005). Prediction of retail product and trimmable fat yields from the four primal cuts in beef cattle using ultrasound or carcass data. J. Anim. Sci., 83: 1353-1360. T e r r y C.A., S a v e l l J.W. (1998). Using ultrasound technology to predict pork carcass composition. J. Anim. Sci., 67: 1279 1284. T y r a M., S z y n d l e r - N ę d z a M., E c k e r t R. (2010 a). Possibilities of using ultrasonography in breeding work with pigs. Part I Analysis of ultrasonic, ultrasonographic and dissection measurements of the most numerous breeds of pigs raised in Poland. Ann. Anim. Sci., 11, 1: 27 40. T y r a M., S z y n d l e r - N ę d z a M., E c k e r t R. (2010 b). Possibilities of using ultrasonography in breeding work with pigs. Part II Relationships between measurements obtained by different techniques and detailed dissection results. Ann. Anim. Sci., 11, 2: 193 205. Accepted for printing 15 III 2011
Using ultrasonography in breeding work with pigs. Part III 369 Mirosław Tyra, Magdalena Szyndler-Nędza, Robert Eckert Możliwości zastosowania techniki ultrasonograficznej (USG) w pracy hodowlanej nad trzodą chlewną. Cz. III Szacowanie zawartości mięsa w tuszy przy użyciu równań regresji opracowanych na podstawie pomiarów ultrasonograficznych Streszczenie Celem badań było opracowanie równań regresji do szacowania procentowej zawartości mięsa w tuszy i ocena ich przydatności. Ponadto, zbadano zależności (korelacje) pomiędzy mięsnością szacowaną różnymi metodami (technikami pomiarowymi) w stosunku do rzeczywistej mięsności uzyskanej z dysekcji szczegółowej. Materiałem badawczym było 476 loszek następujących ras: wbp (151 szt.), pbz (149 szt.), Pietrain (84 szt.) i Duroc (92 szt.), ocenianych w stacjach kontroli (SKURTCh) i ubijanych po osiągnięciu masy ciała 100 kg. Analizowanymi technikami pomiarowymi były: pomiary ultradźwiękowe stosowane w ocenie przyżyciowej (Piglog 105), pomiary ultrasonograficzne (Aloka SSD 500 z sondą liniową), dokonywane przyżyciowo i poubojowo. Punktem odniesienia dla tych grup pomiarów były pomiary dysekcyjne (pomiary wykonywane suwmiarką i planimetrem) oraz wyniki dysekcji szczegółowej określające masę mięsa i tłuszczu oraz ich procentowy udział w tuszy. Najwyższą dokładność opracowanych równań obarczonych niskimi błędami poniżej 2,2% dla pomiarów USG poubojowych i poniżej 2,7% dla pomiarów USG przyżyciowych uzyskano wykorzystując w równaniu pomiary USG: grubość słoniny w punkcie P4, powierzchnię tłuszczu oraz wysokość oka polędwicy. Analiza porównawcza czterech różnych metod szacowania mięsności w stosunku do rzeczywistej mięsności ocenianej na podstawie dysekcji szczegółowej ustaliła następujący ich grading (kolejność): równanie stacyjne (r P = 0,803), równanie oparte na pomiarach USG (Aloka) dokonywanych poubojowo (r P = 0,788), równanie oparte na pomiarach USG (Aloka) dokonywanych przyżyciowo (r P = 0,762) i równanie na podstawie punktów pomiarowych dokonywanych metodą przyżyciową (Piglog; r P = 0,758).