Accuracy of real time ultrasound systems for fat thickness estimation in live cattle.

Abstract

178 Proc. Aust. Soc. Anim. Prod. Vol. 17 ACCURACY OF REAL-TIME ULTRASOUND SYSTEMS FOR FAT THICKNESS ESTIMATION IN LIVE CATTLE R.W. DICKER*, D.G. FOWLER*, D. PERRY# and B. SUNDSTROM+ SUMMARY Ultrasound scanning is an objective method for estimating subcutaneous fat thickness (SFT) in live cattle. It is particularly useful in beef cattle genotype evaluation and marketability research. Estimates of SFT at 12/13th rib and P8 rump sites were made with two real-time imaging ultrasound systems and one echo-ranging system on live steers. Relative accuracies of these systems were compared using regressions of ultrasound estimates on carcase fat measurements, together with frequency distributions of the differences between estimates and measurements. The real-time systems had similar levels of accuracy to the echo-ranging system (residual standard deviations ranged from 1.0 to 2.5 mm), provided more detailed displays of tissue interfaces, but were less portable. Care was needed for the correct location of scanning sites and interpretation of interfaces. Keywords: ultrasound scanning, beef cattle, fat thickness, accuracy. INTRODUCTION Fatness is an important specification for the description of beef cattle. It is one of the characteristics which determine the yield, quality and market suitability of carcases. The most common methods for estimating subcutaneous fat thickness (SFT) in live cattle are visual and manual assessments. However, these methods are subjective and require experienced assessors. The accuracy achieved is not sufficient for most experimental work (Kempster et al. 1982). Ultrasound scanning.provides an objective method for Several systems have been evaluated for use on live cattle Thwaites 1984, Upton et al. 1984, and Sundstrom 1986). In the accuracies of estimates of SFT on live cattle made by echo-ranging system and by two real-time imaging systems. MATERIALS AND METHODS 'Live made at the of yearling in a carcase fattened in cattle ultrasound estimates and carcase measurements of SFT were commonly used standard 12/13th rib and P8 rump siteson two groups, steers. The first group comprised 133 mixed breed' steers judged competition. The second group comprised 31 ye&ling Angus steers a feedlot. Where necessary, steers were restrained in a crush. rib site was located at a point 2/3rd s the width of the M. the midline by palpation. The.P8 rump site was located. at of a line.parallel to the spine from the tuber ischium (pin perpendicular to it from the spinus process of the third Vegetable oil was applied liberally at each scanning site. _ estimating (Tulloh et this paper a currently SFT. al. 1973, we compare used The 12113th longissimus from the intersection bone) and a line sacral vertebra: * N.S.W. Department of Agriculture,~ Agricultural Research and Advisory Station, Glen Innes, N.S.W. 2370 , # N.S.W. Department of Agriculture, Agricultural'Research Centre, Trangie, N.S.W. 2823. + N.S.W. Department of Agriculture, Regional Office, Armidale, N.S.W. 2350. Proc. Aust. Soc. Anim. Prod. VoZ. 17 The echo-ranging ultrasound system, the 'Scanoprobe II' (Sp) [model 731C, with 2MHz transducer frequency, Ithaca Inc, Ithaca:, New York] was used by a highly experienced operator in group 1 and by a less experienced operator in group 2. The two real-time imaging ultrasound systems, the 'Vetscan' (Vs) [3,5MHz, Fischer, Edinburgh] and the 'Vu-300' (Vu) [5MHz, Solus, Sydney], were compared with the Sp in group 2. The Vs was used by an experienced operator and the Vu was used by an inexperienced operator. 179 Carcase fat thickness at the appropriate sites was determined in the chiller by cut and measure techniques. Measurers were experienced, but different for each group. Accuracies of estimates made with each ultrasound system were compared using the regression constants, regression coefficients, residual standard deviations (standard errors of estimate Y) and correlation coefficients for the regressions of ultrasound estimates (Y) on carcase measurements (X). Frequency distributions of differences between estimates and measurements (Y-X) were also used to compare accuracies of systems usedingroup 2. RESULTS Mean (range 2-11 sites. The and lo.1 mm carcase measurements of SFT for the steers in group 1 were 6.5 mm mm) at the 12/13th rib and 9.4 mm (range 3-12 mm) at the P8 rump corresponding means for steers in group 2 were 7.2 mm (range 5-11 mm) (range 7-13 mm). and high * the site Table 1 summarises the statistics for regression of live cattle ultrasound estimates of SFT on carcase measurements of SFT. At both rib rump sites in group 1, low regression constantsand RSD s together with coefficients of regression and correlation indicated high accuracy for estimatesmade by the Sp. At the rib site in group 2,'RSD s for the Sp, Vs and the Vu were similar to those for the Sp in group 1. At the rump in group 2,.however, they were much higher. At both sites, correlation coefficients were lower in group 2 than in group 1. . Table 1 Regression constants (a), 'regression coefficients (b), residual standard deviations (RSD) and correlation coefficients (r) for regressions of live animal ultrasound estimates (Y mm) on carcase measurements (X mm) of subcutaneous fat thickness At the 12/13th rib and P8 rump sites in two groups of steers. Frequency distributionsofthe differences between ultrasound estimates and carcase measurements from group 2 are shown in Table 2. At the 12/13th rib site, 180 77% of carcase a much carcase estimates made measurement. wider range. measurement. . Proc. Aust. Soc. Anim. Prod. Vol. 17 by each of the ultrasound systems were within 1 mm of the However, at the P8 rump site, estimates by all systems had Less than 52% of these estimates were within 1 mm of the About 60% exceeded carcase measurements by up to 5 mm. Table 2 Frequency distributionsof differences between live cattle ultrasound estimates and carcase measurements of subcutaneous fat thickness at the 12/13th rib and P8 rump sites on the 31 steers from group 2 (+ve value is an overestimate) DISCUSSION Relative accuracy of each system was indicated not only by high correlation coefficients for regressions of scan estimates on carcase measurements, but also by low RSD s for predictions of scan estimates, low regression constants and regression coefficients which were close to 1 (Tulloh et al. 1973, Kempster et al. 1982 and,Thwaites 1984). RSD s from the regression were a measure of the precision of the relationship. Correlation coefficients were a measure of the intensity or consistency of association, but could be strongly influenced by the . number of estimates made and the range of SFT s measured.. ' A high- level of accuracy was demonstrated for estimates of SFT made with the echo-ranging system when it was used atboth 12/13th rib and P8 rump sites in `group 1. Similar precision, as indicated by RSD s, was 'also obtained with this system, and with the two real-time imaging systems, at the rib:site in group 2. The low correlation coefficients for systems used in group 2 could have been due to the smaller number of steers, the narrower range of SFT's, and possibly the generally lower levelsof operator experience in this group than in group 1. When compared directly in group 2, each of the ultrasound systems accurately estimated SFT at the 12/13th rib site. Our RSD s were lower than those reported for the simple mechanically coupled system, the 'Scanogram' (Ithaca Inc., Ithaca, New York), by Tulloh et al. (1973) and for the 'Scanoprobe' (Ithaca Inc., Ithaca, New York) by Upton et al. (1984).. At theLP8 rump site, 'hoyever, accuracy of all systems was low. Here, high RSD s indicated a lower level of precision than at the rib site, and high regression constants indicated that each system was uniformly over-estimating SFT on the rump. As seen from the frequency distributions of differences between scans and measurements, most estimates made by each system at the rib site were not Proc. Aust. Soc. Anim. Prod. Vol. 17 181 more than 1 mm above or below the carcase measurement. At the rump site, however, many estimates were up to 5 mm above the carcase measurement. This overestimation of SFT on the rump could have been due to trimming-knife damage in the 'rumping-out' process prior to hide-puller attachment on the slaughter floor. However, we thought that it was more likely a result of mislocation of the P8 site on the carcase by measurers in the chiller. As well as predictive ability, other factors listed by Thwaites (1984) for consideration,when evaluating ultrasonic systems under field conditions,included ease of operation and portability. A feature in the operation of the real-time imaging systems was the detailed screen display of a cross-section of the skin, the subcutaneous fat and the underlying muscle. Rapidly changing contours of these tissue interfaces were seen, particularly near the P8 rump site, as the transducer was moved across the hide of the animal. SFT at rib or rump sites was estimated from a digital display of the distance between cursors set on appropriate skin/fat/muscle interfaces by the operator. The Vu, with its higher frequency transducer, had a greater axial was ' resolution and provided a more detailed screen display than the Vs. Care needed to correctly interpret interfaces and avoid inconsistent estimates. With the echo-ranging system, these interfaces were not as visible. Light-emitting diodes provided an all or none indication of tissue interfaces, and some experience was needed by the operator to correctly read the estimate of SFT. The two real-time imaging systems were desk-top units with a keyboard and video display screen. They required permanent location next to the cattle 'crush, an external power source and protection from weather and glare. On the other hand, the echo-ranging system was a small self-contained, battery-powered unit, ,easily carried between animals by the operator. Ultrasound scanning systems have particular application to beef cattle genotype evaluation and marketability research. This comparison illustrated the use of two systems for estimating SFT inlive cattle. The real-time systems proved to be as accurate as the commonly used echo-ranging system, provided more detailed screen displays of tissue interfaces,but were also more costly. ACKNOWLEDGEMENTS stance with the Research and REFERENCES KEMPSTER, A.J., CUTHBERTSON, A. and HARRINGTON, G. (1982). 'Carcase Evaluation Livestock Breeding, Production and Marketing' (Granada: London). SUNDSTROM, B. (1986). Report to Ausmeat - November 1986. 'Agriculture, New South Wales). THWAITES, C.J. (1984). AMRC Review. 2:23 TULLOH,. N.M., .TRUSCOTT, T.G. and LANG, C.P. (1973). Report to Australian Meat Board - May 1973. (School of Agriculture and Forestry, University of Melbourne). UPTON I W.H ., RYAN, D.M. , MANSFIELD, B.W. and SUNDSTROM, B. 15:764 Aust. Soc. 'Anim. Prod. (1984). Proc. . (Department of