| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ANIMAL PRODUCTS |
* Institute of Animal Breeding and Husbandry with Veterinary Clinic, Faculty of Agriculture, Martin-Luther-University, Halle, D-06108 Germany;
and
Faculty of Chemistry, University of Bielefeld, Bielefeld, D-33615 Germany; and
and
Milk Recording Association of Saxony-Anhalt, Halle, D-06118 Germany
Abstract
The objective of this study was to compare impedance spectroscopy with resistance measurements at a single frequency (50 kHz) for the prediction of lamb carcass composition. The impedance spectrum is usually recorded by measuring the complex impedance at various frequencies (frequency domain); however, in this study, we also applied the faster and simpler measurement in the time domain (application of a current step and measurement of the voltage response). The study was carried out on 24 male, German Blackheaded Mutton lambs with an average BW of 45 kg. Frequency- and time domain-based impedance measurements were collected at 20 min and 24 h postmortem with different electrode placements. Real and imaginary parts at various frequencies were calculated from the locus diagram. Left sides were dissected into lean, fat, and bone, and right sides were ground to determine actual carcass composition. Crude fat, crude protein, and moisture were chemically analyzed on ground samples. Frequency- and time domain-based measurements did not provide the same absolute impedance values; however, the high correlations (P < 0.001) between these methods for the "real parts" showed that they ranked individuals in the same order. Most of the time domain data correlated higher to carcass composition than did the frequency domain data. The real parts of impedance showed correlations between -0.37 (P > 0.05) and -0.74 (P < 0.001) to water, crude fat, lean, and fatty tissue, whereas the relations to CP were much lower (from 0.00 to -0.47, P < 0.05). Electrode placements at different locations did not substantially improve the correlations with carcass composition. The "imaginary parts" of impedance were not suitable for the prediction of carcass composition. The highest accuracy (R2 = 0.66) was reached for the estimation of crude fat percentage by a regression equation with the time domain-based impedance measured at 24 h postmortem. Furthermore, there was not a clear superiority of measurements in a wide frequency range over a single frequency measurement at 50 kHz for the prediction of carcass composition. Even though we calculated the impedance at 50 kHz based on the locus diagram, which allowed for a high precision for predicting this impedance trait, single-frequency impedance devices typically used in practice cannot record the locus diagram and, therefore, exhibit a greater amount of uncertainty.
Key Words: Carcass Composition • Impedance • Lamb
Introduction
Breeding strategies for improving carcass quality require a precise estimation of lean and fat content. Impedance measurement is a fast, nondestructive technique and only requires fairly simple equipment. It can predict the volume of body compartments involved in current flow (extracellular fluid at low frequencies and total body water at higher frequencies). The estimation of body composition should be possible if a constant water content of body compartments is assumed. Several authors reported results at 50 kHz in farm animals (Berg et al., 1996
; Swantek et al., 1999; Velazco et al., 1999) and found medium to high correlations with the carcass composition. Schoeller et al. (2000) assumed that the impedance at 50 kHz is overwhelmed by the extracellular water, and the high correlation between impedance and the total body water, as well as the fat free mass, was due to the tight relation of these compartments to the extracellular water (at least in healthy humans).
Because the impedance is frequency dependent, it is favorable to extend the frequency range for greater accuracy. Research on humans at various frequencies suggested confounding frequencies for the best prediction of body composition (Pietrobelli et al., 1998; Tagliabue et al, 2001). Moreover, measurements at direct current or very low frequencies cause further technical problems (Pliquett et al., 2003).
With this in mind, we suggest use of the impedance at direct current (R0) and at a very high frequency (R
), extrapolated from the impedance spectrum. These two parameters are independent of the measurement procedure and do not rely on a compromise with respect to the frequency range. Therefore, we investigated the suitability of measurements in a wide frequency range (impedance spectroscopy) for prediction of lamb carcass composition compared to the impedance at 50 kHz. In addition to the commonly used impedance measurement with the variation of frequencies, we applied another faster and simpler method.
Materials and Methods
Animals
With respect to the geometrical dependence of impedance, we used lambs with a nearly balanced body shape. Twenty-four intact male lambs of the same breed (German Blackheaded Mutton) within a narrow range for live weight were included in the present study. Sheep were weaned at 3 mo of age and had ad libitum access to a concentrate-hay diet. The concentrate contained 19.0% CP, 4.4% crude fat, 10.3% crude fiber, and 10.2 MJ of ME (DM basis). Sheep were weighed weekly, and feed was withheld for 24 h before slaughter at a final BW of 44 to 46 kg. Lambs were slaughtered according to the humane practices of normal industrial procedures, and carcasses were chilled at 2°C for 24 h.
Impedance Measurements
Biological tissues consist of cells surrounded by an insulating plasma membrane. This membrane, contacted by conductive electrolytes, behaves like a capacitor. Direct current (dc) does not cross the cell membranes but causes a transient displacement current due to the membrane charging. Alternating current (ac) crosses the membrane due to the periodical displacement current. The resistance of the membrane (Rm) is inversely proportional to the frequency. Membrane resistance is calculated as Rm = 1/(2
fCm), where f is the frequency and Cm the lumped capacity of the cell membranes. This current is called "imaginary" because no net transport of charge carriers (ions) occurs, which is opposite from "real" ion transport in an aqueous electrolyte solution. Membranes exhibit a high resistance at low frequencies (<100 Hz) so that most of the current flows through the extracellular electrolytes. With increasing frequency, more and more current crosses the cell membrane (displacement current) and, thus, increases the imaginary current through the tissue. At the characteristic frequency, the magnitude of the imaginary part becomes a maximum and disappears at very high frequencies when the resistance of the membrane approaches zero (Foster and Schwan, 1989). The fat-free mass can be predicted by impedance because of the greater electrolyte content of lean compared to adipose tissue or bone.
We achieved the measurement of the electrical impedance in two general ways: frequency- and time-domain approaches. The following impedance traits were calculated from the locus diagram: R0 = real part at dc; R = real part at a very high frequency; Rfc = real part at the characteristic frequency; Xfc = imaginary part at the characteristic frequency; Zfc = impedance at the characteristic frequency 
; R50 = real part at 50 kHz; X50= imaginary part at 50 kHz; and Z50 = impedance at 50 kHz 
(refer to the appendix herein and to Pliquett et al., 2003).
Devices for Impedance Measurement
The four-electrode interface, used in the experiments, was developed in our laboratories. Needle electrodes (syringe, diameter 0.95 mm, length 35 mm) were mounted directly onto an amplifier matched to 50 
. Therefore, the length of the cables between the electrodes and the impedance measurement device (1.5 m) was not important. To keep a good signal-to-noise ratio (SNR), one of the current electrodes carried the current amplifier, whereas the other current electrode had an internal pull-down resistor of 1 k. The voltage electrodes were both attached to the buffer amplifiers. The time domain-based measurement system, consisting of a rectangular wave generator, a voltage/current converter for the application of the current stimulus (100 µA, 2 kHz rectangular wave), and an instrumentation amplifier for monitoring the voltage across the sensor electrodes were also developed in our laboratories. A microcontroller generated the stimulus and a digital oscilloscope THS720 (Tektronix, Beaverton, OR) recorded the current and voltage trace. The data were transferred to a laptop computer for further processing. For comparison with experiments with frequency domain-based measurements reported in literature, a Solartron 1260 gain phase amplifier (Schlumberger. Ltd., Kingston, U.K.) was used, and was connected via a switch and a matching amplifier to the same set of electrodes.
Electrode Placements
The impedance of carcasses was measured at 20 min and 24 h postmortem. The following configurations of electrode placement were examined (Figure 1): Ventral I = current electrodes in the right flexor digitorum close to the carpal condyle and in the left semimembranosus close to the tuberositas tibiae, as well as voltage electrodes in the right pectoralis superficialis close to the elbow joint and in the left semimembranosus close to the condylus lateralis; Ventral II (only at 24 h postmortem) = a simple contralateral switching to Ventral I; and Dorsal = current electrodes 3 cm from the dorsal midline at the first coccygeal vertebra and 5 cm cranial from the tip of the shoulder blade, with voltage electrodes 3 cm from the dorsal midline, 5 cm cranial of first coccygeal vertebra, and at the tip of the shoulder blade. Three impedance measurements were carried out for every variation. The electrodes were newly placed after every measurement. Additionally, the distance of detector electrodes, carcass temperature, and carcass weight were recorded.
|
Carcasses were split along the dorsal midline after the impedance measurements 24 h postmortem. Right sides were ground once in a mill (H35 S/2, Condux Mahltechnik GmbH, Hanau, Germany), and twice in a mincer (W130, SPOMASZ, Poland) with a 3-mm plate. A 200-g sample was collected for chemical analysis. Water content was ascertained by drying at 105°C, crude fat by n-hexane extraction (Soxtec HT2, TECATOR, Uppsala, Sweden), and CP according to the Kjeldahl procedure (Vapodest 45, Gerhardt Apparate GmbH & Co. KG, Bonn, Germany). Percentages of protein, fat, and moisture were used for assessment of the right side composition. Furthermore, the content of fat-free soft tissue was calculated using the equation of Jenkins et al. (1988). Left sides were fabricated into wholesale cuts without trimming according to the guidelines of the German Agricultural Society (DLG: Scheper and Scholz, 1985), and than frozen at -20°C. Wholesale cuts were then dissected into lean, fat, and bones after thawing, with thawing loss added to the lean portion. Tissue masses were summed, and lean and fat yields were determined as a percentage of the left side weight.
Statistical Methods
For the statistical analysis, means of repeated impedance measurements were used: three replicates for 20 min postmortem dorsal and ventral (I) electrode placement, three replicates for 24 h postmortem dorsal electrode placement, and six replicates for ventral (I and II) electrode placement. Pearson correlations between frequency- and time domain-based impedance data were calculated using PROC CORR of SAS (SAS Inst., Inc., Cary, NC). The three impedance measurements for every electrode placement, as well as the results for ventral I and II, were averaged for the prediction of carcass composition. Partial correlations between impedance and carcass composition with electrode distance and carcass weight as covariates were calculated because a certain variation in body shape still exists in spite of similar BW, breed, and gender. Furthermore, the data were analyzed by linear regression procedures, and impedance data, electrode distance, and carcass weight served as independent variables. Stepwise forward regression was used to determine significant variables, and only variables with F < 0.05 were included in the equation.
Results and Discussion
Carcass Data
Carcass traits of sheep used in the present study are shown in Table 1. Even though weight, breed, and gender were similar, carcass composition varied considerably. Lean meat percentage varied between 56.2 and 65.2%, and the fat content varied between 14.7 and 27.7%.
|
The impedance measurement in the frequency domain is the method of choice for the prediction of body composition in humans and animals by impedance spectroscopy because time domain impedance devices are not available commercially. It is recognized to be very accurate because of the use of selective amplifiers that greatly reduce the noise of the measurement. Another approach is to use time domain-based measurements. The advantage of this method is to use a less-complex apparatus and a faster measurement compared to the frequency domain method, which is important for practical use on live animals. However, the broad bandwidth amplifier for recording the response is always associated with higher noise. The passive electrical properties of linear and time-invariant objects are exactly the same when measured in time and frequency domain (Pliquett et al., 2000). The term "linear" in this study means that the resistance (R = U/I; where U = voltage and I = current intensity) is constant and not a function of the voltage as in nonlinear objects. Time invariance means no changes of the object during the time of the measurement. Because biological objects show slightly different results in comparison with technical objects, we compared the results of both measurements.
In our experiment, the magnitude of impedance was lower (P < 0.001) when measured in the time domain compared with frequency domain (Table 2). The lowest impedance values were measured by both methods 20 min postmortem with dorsal electrode placement, and values were higher at ventral placement due to the greater electrode distance. The impedance increased as a result of decreasing tissue temperature 24 h postmortem.
|
. In this case we did not use the averages of repeated measurements. The data were calculated on the basis of single measurements of all electrode placements. The correlations between time and frequency domains for the real parts and Z were high, and the results differed by a factor of 1.1 to 1.6. This means that both methods do not provide identical absolute values, but they are ranking individual carcasses in the same order. The imaginary parts corresponded even less, with values from the time domain of 2.5- to 4.7-fold lower than from frequency domain. The correlations for the imaginary parts were much lower than for the real parts.
|
; Berg et al., 1997), our values for R50 and X50 were higher. This is due to different electrode placement, which resulted in a 10- to 20-cm greater distance between the voltage electrodes. Berg and Marchello (1994) carried out impedance measurements on chilled lamb carcasses with nearly the same electrode distance as employed in our study and presented R50- and X50-values in the same range as our time domain data. Correlations Between the Carcass Traits
The different compartments of carcass were highly (P < 0.001) correlated among each other (Table 4). Water content was equivalent to total body water and was highly correlated (r > 0.80) with other compartments (except CP). A reasonable estimation of lean and fat content would be indirectly possible, if the prediction of total body water by impedance succeeds exactly.
|
Most of the time domain data correlated higher with carcass composition than frequency domain data. Impedance 20 min (Table 5) and 24 h postmortem (Table 6) showed medium to high correlations with water, crude fat, and fat-free soft tissue. In contrast, the relations with CP were much lower. Thomson et al. (1997) found, by multifrequency impedance measurements in beef, lower correlations with carcass protein than with the water content. In vivo measurements at 50 kHz in sows by Kraetzl et al. (1995) also resulted in lower correlations for protein than for water and fat.
|
|
at 24 h postmortem with the water content had approximately the same level for the frequency domain-based measurements. In the time domain, R0 was more strongly correlated with the water content than R. Evidently, there is a close relationship between the extracellular and total water of carcass, so that an indirect prediction of total water by R0 is understandable. Lean and fat percentages were more highly correlated with R than with R0, with the exception of measurements in the time domain 24 h postmortem. Oliver et al. (2001) used multifrequency impedance in hams and found higher correlations between fat thickness and R. If the difference between R0 and R is used, the correlations with carcass composition decreases (data not shown), and further reductions in correlations were observed for the quotient Py = (R0 - R)/R0 with carcass composition. Pliquett et al. (1995) recommend this quotient for the detection of PSE meat in pigs because it has the advantage of being independent from geometry and temperature.
Correlations of Rfc and R50 with carcass composition differed little from each other at 20 min postmortem. Stronger relationships for R50 than for Rfc were observed 24 h postmortem at all electrode placements. In lambs of different breeds and weights, Berg et al. (1997) reported a correlation of -0.32 between R50 at 24 h postmortem and lean percentage, whereas Cosgrove et al. (1988) found a correlation of -0.79 between R50 and lean percentage. Our values ranged from -0.46 to -0.73. The suitability of impedance at fc and 50 kHz for the prediction of carcass composition is questioned in the literature. The total body water in men (Ward and Stroud, 2001) and rats (Cornish et al., 1993) can be estimated with a higher precision by measurements at fc. On the other hand, the water, fat, and protein compositions of lamb carcasses were correlated similarly with the impedance at fc and at 50 kHz in studies by Hegarty et al. (1998).
The imaginary parts (Xfc and X50) do not seem to be suitable for the prediction of carcass composition. These imaginary parts had much lower correlations than the real parts. This is confirmed by other studies on lambs, pigs, and turkeys (Grimes et al., 1990; Berg and Marchello, 1994; Marchello et al., 1999).
Due to the small variation between correlations of different variables, it is impossible to decide which variable is the best to predict carcass composition. Lean and fat percentage were correlated the highest with R in the frequency domain, but no variable can be preferred in the time domain. No superiority of any variable is evident for the prediction of water and crude fat by frequency domain-based impedance. In the time domain, water and crude fat were highly correlated with R0 at 20 min postmortem and with R50 at 24 h postmortem.
Different correlations were found at the various electrode placements. There was a tendency for higher correlations for dorsal placement at 20 min postmortem, but correlations were higher at 24 h postmortem for ventral placement. One must consider that three single measurements were averaged at 20 min postmortem for each electrode placement; however, six single measurements (Ventral I and Ventral II) were considered for ventral placement at 24 h postmortem. The higher correlations were attributed to the higher reproducibility at the ventral placement at 24 h postmortem. The repeatability is generally influenced by the accuracy of the electrode position and the reproducibility of the impedance measurement itself. We even found a variation in the impedance data when the electrodes were already in place (our unpublished data).
Prediction of Carcass Composition by Regression Equations
Predictions of carcass composition by impedance measurements in the frequency and time domains for dorsal and ventral electrode placements are presented in Tables 7 and 8, respectively. We used the dorsal electrode placement at 20 min postmortem and the ventral electrode placement at 24 h postmortem because they had higher correlations with carcass composition. Time domain-based measurements resulted in a higher R2, and lower root mean square error (RMSE), than measurements in the frequency domain. The highest accuracy was reached 20 min postmortem for the fat-free soft tissue (FFST% = 88.340 - 0.278R0 + 0.558L), whereas 24-h-postmortem crude fat could be predicted with the highest precision (Fat% = -39.812 + 1.220W + 0.087Zfc). Berg et al. (1997) reported R2-values for percentage lean of 0.53 and 0.58, which is comparable to our results. Other authors predicted the lean and fat tissue, or chemical compounds in lambs, with greater accuracy, but they included weight in their regression equations (Cosgrove et al. 1988; Berg and Marchello, 1994; Slanger et al., 1994). A comparison with our results is not advisable because we used lambs of similar weight.
|
|
Theoretically, frequency- and time domain-based impedance measurements yield the same result. Both methods are highly correlated, at least for variables relevant for carcass composition. Due to the use of frequency-selective amplifiers, the measurement in the frequency domain exhibits superior signal-to-noise ratio, whereas the technically simpler and faster measurement in the time domain should be preferred for application in practice. Results could only partially confirm the higher accuracy of carcass composition by impedance measurements in a wide frequency range compared with a single frequency of 50 kHz. However, we calculated the impedance at 50 kHz from multifrequency measurements, yielding a higher precision of this value compared with the reading at a single frequency.
APPENDIX
The electrical impedance (Z) is the complex resistance: Z = R + jX. The real part (R) is due to net transport of charge carriers (current flow in an electrolyte), whereas the imaginary part (X) is due to the displacement current at cell membranes (j is the imaginary unit 
; Foster and Schwan, 1989).
Measurements in the Frequency Domain
The impedance is measured successively at various frequencies. The results can be presented at the locus diagram (Figure 2). This is the imaginary part of the impedance vs. the real part. The locus diagram is a depressed semicircle with the low frequencies at the right side. The real part decreases with increasing frequency, whereas the magnitude of the imaginary part increases. The imaginary part reaches a minimum at the characteristic frequency (fc) and vanishes at high frequencies. For example, at direct current (dc), only the extracellular electrolytes conduct the current. With increasing frequency, more and more of the intracellular electrolytes are involved in current flow due to the displacement current across the membranes. A good estimate for the fc is when the resistance of the membranes equals the resistance of the electrolytes. The cell membranes are quasi-shortened at very high frequencies; thus, the intra- and the extracellular electrolytes are involved in current flow, which results in a fairly small real part of the resistance.
|
, when at very high frequencies the resistance of the cell membranes vanishes. Practically, the curve does not reach the X-axis, which hints at additional frequency dispersion at low, as well as high, frequencies. Measurements in the Time Domain
In contrast to the variation of frequencies, only one signal with a broad bandwidth is applied and the response measured. The application of a voltage step at the time (t0) causes the membrane to charge up. However, at t0, no voltage drops across the membrane (i.e., the current (I0) flows through the intra- and extracellular electrolytes). The current decreases exponentially due to the charge up of the membrane. Finally, if the membrane is charged, the current through intracellular fluid becomes zero and all the remaining current flows through the extracellular fluid. The standard method for converting time domain data into the frequency domain is the Fourier transformation (Krylov, 1977).
Comparison of Time and Frequency Domain Measurements
Theoretically, both measurements yield the same result; however, due to the use of frequency-selective amplifiers, the measurement in the frequency domain exhibits superior signal-to-noise ratio. The advantage of the time domain-based measurements is speed. For the measurements of the spectrum between 1 kHz and 1 MHz with 60 discrete frequencies, about 40 s is necessary when measured in the frequency domain but only 1 ms for the same spectrum in time domain.
1 Correspondence: D-06108 Halle, Adam-Kuckhoff-Str. 35 (phone: 0345-5522346; fax: 0345-5527106; e-mail: altmann{at}landw.uni-halle.de).
Received for publication January 13, 2003. Accepted for publication October 16, 2003.
Literature Cited
Berg, E. P., and M. J. Marchello. 1994. Bioelectrical impedance analysis for the prediction of fat-free mass in lambs and lamb carcasses. J. Anim. Sci. 72:322–329.[Abstract]
Berg, E. P., M. K. Neary, J. C. Forrest, D. L. Thomas, and R. G. Kaufmann. 1996. Assessment of lamb carcass composition from live animal measurement of bioelectrical impedance or ultrasonic tissue depths. J. Anim. Sci. 74:2672–2678.[Abstract]
Berg, E. P., M. K. Neary, J. C. Forrest, D. L. Thomas, and R. G. Kaufmann. 1997. Evaluation of electronic technology to assess lamb carcass composition. J. Anim. Sci. 75:2433–2444.
Cornish, B. H., B. J. Thomas, and L. C. Ward. 1993. Improved prediction of extracellular and total body water using impedance loci generated by multiple frequency bioelectrical impedance analysis. Phys. Med. Biol. 38:337–346.[Medline]
Cosgrove, J. R., J. W. B. King, and D. A. Brodie. 1988. A note on the use of impedance measurements for the prediction of carcass composition in lambs. Anim. Prod. 47:311–315.
Foster, K. R., and H. P. Schwan. 1989. Dielectric properties of tissues and biological materials: A critical review. CRC Crit. Rev. Biomed. Eng. 17:25–104.[Medline]
Grimes, J. L., J. F. Ort, V. L. Christensen, and H. R. Ball. 1990. Bioelectrical impedance analysis of the fat-free mass in breeder turkey hens. Poult. Sci. 69:369–377.
Hegarty, R. S., M. J. McPhee, V. H. Oddy, B. J. Thomas, and L. C. Ward. 1998. Prediction of the chemical composition of lamb carcasses from multi-frequency impedance data. Brit. J. Nutr. 79:169–176.[Medline]
Jenkins, T. G., K. A. Leymaster, and L. M. Turlington. 1988. Estimation of fat-free soft tissue in lamb carcasses by use of carcass and resistive impedance measurements. J. Anim. Sci. 66:2174–2179.
Kraetzl, W. D., S. Thiele, J. Tomczak, and J. Kamphues. 1995. Bioelektrische Impedanzanalyse - eine in vivo-Methode zur Schätzung der Körperzusammensetzung von Sauen. Züchtungskunde 67:132–146.
Krylov, V. A. 1977. A handbook of methods of approximate Fourier transformation and inversion of Laplace transformation. Mir Pub., Moscow.
Marchello, M. J., P. T. Berg, P. M. Swantek, and J. E. Tilton. 1999. Predicting live and carcass lean using bioelectrical impedance technology in pigs. Livest. Prod. Sci. 58:151–157.
Oliver, M. A., I. Gobantes, J. Arnau, J. Elvira, P. Riu, N. Grebol, and J. M. Monfort. 2001. Evaluation of the electrical impedance spectroscopy (EIS) equipment for ham meat quality selection. Meat Sci. 58:305–312.
Pietrobelli, A., P. Morini, N. Battistini, G. Chiumello, C. Nunez, and S. B. Heymsfield. 1998. Appendicular skeletal muscle mass: prediction from multiple frequency segmental bioimpedance analysis. Eur. J. Clin. Nutr. 52:507–511.[Medline]
Pliquett, U., M. Altmann, F. Pliquett, and L. Schöberlein. 2003. Py: A parameter for meat quality. Meat Sci. 65:1429–1437.
Pliquett, U., E. Gersing, and F. Pliquett. 2000. Evaluation of fast time domain based impedance measurements on biological tissue. Biomed. Tech. 45:6–13.
Pliquett, F., U. Pliquett, L. Schöberlein, and K. H. Freywald. 1995. Impedanzmessungen zur Charakterisierung der Fleischbeschaffenheit. Fleischwirtschaft 75:496–498.
Scheper, J., and W. Scholz. 1985. DLG-Schnittführung für die Zerlegung der Schlachtkörper von Rind, Kalb, Schwein und Schaf. DLG-Verlag, Frankfurt/Main.
Schoeller, D. A. 2000. Bioelectrical impedance analysis. What does it measure? Ann. NY Acad. Sci. 904:159–162.[Medline]
Slanger, W. D., M. J. Marchello, J. R. Busboom, H. H. Meyer, L. A. Mitchell, W. F. Hendrix, R. R. Mills, and W. D. Warnock. 1994. Predicting total weight of retail-ready lamb cuts from bioelectrical impedance measurements taken at the processing plant. J. Anim. Sci. 72:1467–1474.[Abstract]
Swantek P. M., M. J. Marchello, J. E. Tilton, and J. D. Crenshaw. 1999. Prediction of fat-free mass of pigs from 50 to 130 kg live weight. J. Anim. Sci. 77:893–897.
Tagliabue, A., A. Andreoli, M. Comelli, S. Bertoli, G. Testolen, G. Oriani, and A. De Lorenzo. 2001. Prediction of lean body mass from multifrequency segmental impedance: influence of adiposity. Acta Diabetol. 38:93–97.[Medline]
Thomson, B. C., B. J. Thomas, L. C. Ward, and M. N. Sillence. 1997. Evaluation of multifrequency bioelectrical impedance data for predicting lean tissue mass in beef cattle. Aust. J. Exp. Agr. 37:743–749.
Velazco, J., J. L. Morrill, and K. K. Grunewald. 1999. Utilization of bioelectrical impedance to predict carcass composition of Holstein steers at 3, 6, 9 and 12 months of age. J. Anim. Sci. 77:131–136.
Ward, L. C., and D. B. Stroud. 2001. Is 50 kHz the optimal frequency of measurement in single frequency bioelectrical impedance analysis? Pages 369–372 in Proc. XI. Intern. Conf. On Electrical Bio-Impedance, Oslo, Norway.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
