HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Steiner, R. Articles by Smith, G. C. Search for Related Content PubMed PubMed Citation Articles by Steiner, R. Articles by Smith, G. C.

Accuracy and repeatability of beef carcass longissimus muscle area measurements¹

R. Steiner^*,2, D. J. Vote^*, K. E. Belk^*,3, J. A. Scanga^*, J. W. Wise, J. D. Tatum^* and G. C. Smith^*

^* Department of Animal Sciences, Colorado State University, Fort Collins, 80523-1171 and and USDA Agricultural Marketing Service, Livestock and Seed Program, Livestockand Meat Standardization Branch, Washington, DC 20090-6456

	Abstract

Top Abstract Introduction Materials and Methods Results and Discussion Implications Literature Cited

 
This study was conducted to determine the accuracy and repeatability of beef carcass longissimus muscle area (LMA) measurements obtained by three different methods. Longissimus muscle area for beef carcass sides (n = 100) randomly selected in a commercial packing plant was determined: 1) independently by three USDA grading supervisor "experts" using the grid method to obtain triplicate measurements of the same longissimus muscle (LM); 2) by three different Colorado State University personnel tracing the LMA on acetate paper and subsequently measuring the area via a polar planimeter three different times (total of 3 x 3 = 9 observations/LM); and 3) by use of two identical video image analysis (VIA) instruments making triplicate measurements for each LM using three different procedures. Video image analysis Procedure 1 required that LMA be measured by placing the camera head unit over the LM and collecting three sequential images without moving the camera head unit while carcasses were in a stationary position; Procedure 2 required measurement of LMA by placing the camera head unit over the LM and collecting three images, but removing and repositioning the camera head unit between collection of each image while carcasses were in a stationary position; and Procedure 3 required that LMA be measured by placing the camera head unit over the LM and obtaining an image while carcasses were in continuous motion (chain speed of 360 carcasses/h) during three different circulations past the grading stand. Overall, VIA-derived LMA measurements were highly accurate for all three procedures compared with expert-gridded (R² = 0.92, 0.90, and 0.84 for Procedures 1, 2, and 3, respectively) and acetate/planimeter-traced (R² = 0.94, 0.93, and 0.86 for Procedures 1, 2, and 3, respectively) LMA measurements. Instrument LMA repeatability also was comparable to expert-gridded and acetate/planimeter-traced LMA repeatability, as the means of the absolute differences between individual measurements and the average of those same measurements per LM were 1.29, 1.35, 0.52, 0.84, and 1.87 cm² for expert-gridded, acetate/planimeter-traced, and VIA Procedures 1, 2, and 3, respectively. Therefore, VIA instrumentation can be used to assess beef carcass LMA in both a stationary and operational scenario with high levels of accuracy and repeatability.

Key Words: Beef • Carcass Grading • Instrumentation • Longissimus dorsi

	Introduction

Top Abstract Introduction Materials and Methods Results and Discussion Implications Literature Cited

 
At present, USDA on-line graders visually estimate the extent to which longissimus muscle area (LMA) meets or deviates from a prescribed "expected" value (based on a standard relationship between LMA and hot carcass weight), and then apply an appropriate adjustment to the preliminary yield grade (YG) in computing a final YG. If on-line graders desire greater accuracy in assessing LMA, a plastic grid may be used to obtain an objective area measurement of the longissimus muscle (LM). However, the use of a grid is time consuming and impractical for use in today’s commercial operations, where large numbers of carcasses are evaluated in a short period of time.

Variance associated with use of the plastic grid technique among different individuals obtaining LMA measurements or among different replications of measurement on the same LM is not well documented. Additionally, no data exist regarding the accuracy or repeatability of expert graders in measuring LMA using a plastic grid, making it impossible to quantify comparative expectations of expert graders and instruments with regard to either accuracy or repeatability of their estimates.

Previous studies (Borggaard et al., 1996; Belk et al., 1998; Cannell et al., 1999) have demonstrated instrument assessment of LMA to be highly correlated with expert LMA measurements. However, no studies have been conducted to determine the repeatability of instrument measures of LMA. Consequently, a baseline for accuracy and repeatability of LMA measurement is needed if instruments are to be introduced for commercial use in augmenting the application of USDA YG. The objective of this study was to determine the accuracy and repeatability of LMA measurements obtained by expert USDA graders using the plastic grid method, tracing the LM and subsequently measuring the LMA using a compensating polar planimeter, and by video image analysis (VIA) instrumentation.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results and Discussion Implications Literature Cited

 
This study determined the accuracy and repeatability of LMA measurements attained by: 1) a panel of expert USDA graders using a plastic grid (expert-gridded); 2) acetate tracings subsequently measured using a compensating polar planimeter (acetate/planimeter-traced); and 3) two VIA instruments (Computer Vision System Cold Camera; Research Management Systems, USA Inc., Fort Collins, CO).

Fifty carcasses were selected from normal production in a commercial packing plant (Excel Corp., Fort Morgan, CO) to exhibit as much variability in weight and LMA as was possible. Both sides of the carcass (n = 100) were used to collect independent LMA measurements. Selected carcasses were placed on static rails and ribbed between the 12th and 13th ribs to expose the LM immediately before data collection.

Following carcass selection and ribbing, three "experts" (employees of the USDA-Agricultural Marketing Service), separately and independently gridded the LM of each carcass side. Each independent determination of LMA via the plastic grid (clear plastic sheet with small dots arranged at the center of a square 0.645 cm² in area; Art Services, Washington, DC) by experts was recorded by personnel of Colorado State University in a manner that maintained the confidentiality of each recorded measurement. In addition, independent tracings of each exposed LM were obtained by three trained Colorado State University personnel using acetate paper according to the procedures of Naumann et al. (1951). After completion of data collection, acetate paper tracings were transported to Colorado State University, where each set of three tracings for each LM was measured using a compensating polar planimeter, again by three different personnel (a total of 3 x 3 = 9 planimeter measures per LM), to determine acetate/planimeter-traced LMA.

Two portable VIA instruments were operated by two differing designates (one for each instrument) of Research Management Systems USA, Inc. Video image analysis LMA measurement accuracy and repeatability were determined via three distinct procedures in order to establish both instrument performance standards (Procedures 1 and 2) and operational performance standards (Procedure 3).

Procedure 1—Triple Trigger
Longissimus muscle area was measured with VIA instruments in Procedure 1 by properly placing the camera head unit (with as much precision as was possible for each side) over each LM and obtaining three sequential but separate images without moving the camera head unit (300 observations for each of the two different portable VIA instruments). Images collected by each portable VIA instrument were obtained while the carcasses were in a stationary position.

Procedure 2—Placement
Longissimus muscle area was measured with VIA instruments in Procedure 2 by placing the camera head unit with as much precision as was possible over the LM of each carcass side, obtaining one image, removing the camera head unit, and repeating this process two more times (300 total observations for each of the two VIA instruments). Images were obtained while carcasses were in a stationary position.

To obtain VIA instrument measurements of LMA for the stationary procedures outlined above, each carcass was evaluated immediately following carcass ribbing in the regrade bay of the "sales cooler" before any carcass movement occurred.

Procedure 3—Operational
Longissimus muscle area was measured with VIA instruments in Procedure 3 by circulating carcasses in continuous motion (chain speed of 360 carcasses/h) past the grading stand three separate times. As the carcasses were transported past the two different portable VIA instruments during each circulation, the camera head units of each instrument were placed over the LM and one image was obtained (300 total observations per portable VIA instrument). After the completion of each circulation, experts again, independently, measured via the plastic grid the LMA of each carcass side.

Statistical Analysis.
The simple average (arithmetic mean) of the three independent measurements of the LMA of a carcass side obtained by experts and the average of the three independent tracings (and subsequent measurement, three times per tracing, with a compensating polar planimeter) were defined as the "actual" measures of LMA. Accuracy of instrument estimates of the LMA for a given carcass side was assessed by comparing each instrument estimate to the "actual" LMA measures for that carcass side and determining the strength of the relationships by using the REG procedure of SAS (SAS Inst., Inc., Cary, NC).

Repeatability of VIA estimates of LMA was assessed first by determining the mean of the absolute differences between individual VIA LMA measurements and the average of those same measurements in each set of three images obtained per instrument for the three procedures of the study. These mean absolute differences were compared to the same computations made among LMA measurements of the three experts using the plastic grid method. In addition, VIA mean absolute differences were compared (for Procedures 1, 2, and 3) to the same computations for acetate tracing and subsequent measurement with a compensating polar planimeter (nine measures per LM).

Secondly, repeatability for expert-gridded, acetate/planimeter-traced, and VIA-measured LMA (by procedure) was determined by evaluating variance components using the MIXED models procedures of SAS according to Montgomery (1997). Variance components estimation was conducted using REML. For each LM, the proportion of total variance due to measurement replication was calculated as (. The model used to partition variance was:

where Y_ijk = ijkth measured LMA, µ = overall mean, _i = random effect of the ith operator or differing VIA instruments, ß_j = random effect of the jth carcass side, and _ijk = residual error.

Means were separated using the TTEST procedure (paired comparisons t-test) of SAS in order to determine if differences existed between the left and right sides of the carcasses. Mean absolute differences between cameras, by procedure, were tested using the TTEST (Cochran adjustment for unequal variances) procedure of SAS.

	Results and Discussion

Top Abstract Introduction Materials and Methods Results and Discussion Implications Literature Cited

 
Carcasses in the sample population (n = 50) had a mean overall maturity of A⁶⁸, USDA yield grade of 3.1, fat thickness of 1.1 cm, estimated percentage of kidney, pelvic, and heart fat of 2.1, and a mean hot carcass weight of 366.1 kg. Thus, it was apparent that the sample population used in this study closely approximated the carcass trait means reported in the National Beef Quality Audit—2000 (Mckenna et al., 2002).

Means, standard deviations, minimal and maximal values for LMA measurements are presented in Table 1 by carcass side and measurement method. The average LMA of the left side was larger (P < 0.05) than the average area of the right side for expert-gridded, acetate/planimeter-traced, and VIA Procedures 1 and 2. Considerable variation in beef carcass LMA between the left and right sides has also been reported by Butler et al. (1956), Cole et al. (1960), and Carpenter et al. (1961). Hedrick et al. (1965) indicated that differences in LMA between the left and right sides of the same carcass can occur if the sides are not ribbed at the same angle. Results suggested that considerable variation existed in ribbing techniques between the left and right sides of carcasses used in this study.

It was noteworthy that VIA-derived LMA, for all three procedures, explained more of the observed variability in the acetate/planimeter-traced method than it did in the expert-gridded LMA method. Overall VIA accuracy for Procedures 1, 2, and 3 tended to be 2 to 3% higher (R² x 100) when compared to the acetate/planimeter-traced method vs. the expert-gridded method. The difference in VIA-derived LMA accuracy between the acetate/planimeter method and the expert-gridded method was most likely due to the similarity between how VIA and acetate/planimeter-traced measurements were obtained; both the VIA and acetate/planimeter-traced methods trace the outline of the LM, which is then used to obtain the area. In contrast, the expert-gridded method uses a different approach to measure LMA; an operator counts the number of squares (0.1 in², converted to metric) that cover the LM, depending on whether the dot at the center of each square appears to touch the LM or not, to determine the area. In the present study, the use of the grid by expert USDA graders resulted in slightly larger LMA measurements than did acetate/planimeter-traced readings, which is in agreement with results reported by Henderson et al. (1966).

Frequency distributions of differences between VIA-derived LMA (combining both VIA instrument measurements) and mean expert-gridded LMA or mean acetate/planimeter-traced LMA are displayed in Figure 1 by procedure. Frequency distributions of differences between VIA-derived LMA and acetate/planimeter-traced LMA were almost perfectly centered on zero, whereas the frequency distributions of differences between VIA-derived LMA and expert-gridded LMA were shifted to the right for each procedure. Additionally, differences between VIA-measured LMA and acetate/planimeter-traced LMA, and VIA-measured LMA and expert-gridded LMA differed (P < 0.05). Thus, experts tended, using a plastic grid, to measure LMA larger than when LMA was determined using the acetate/planimeter-traced method, making it look like the VIA measurements of LMA were too small, when in fact that was not the case. Therefore, the acetate/planimeter-traced technique of measuring LMA may provide a more accurate representation of actual LMA, and should be the technique used to determine the accuracy of any instrument-derived LMA. This conclusion was substantiated by the findings of Bodwell et al. (1959), who, in a review of error introduced in the measurement of LMA by planimeter and grid, indicated that the LMA of beef carcasses was accurately measured by tracing the boundary and measuring the tracing with a planimeter. Regardless of what is considered to be the actual measurement of LMA, if there is a consistent shift in an instrument’s measurements compared with the actual, then the instrument’s measurements can be adjusted by adding or subtracting a constant value to reflect the difference observed between the two measurement methods. Based on the data from the current study, if the grid method was defined as the actual measurement of LMA, then it may be useful to add a constant value of 5 to 6 cm² to each of the VIA measurements and 4 cm² to each of the acetate/planimeter tracing measurements to more closely approximate the defined actual value of LMA. If the acetate/planimeter-traced method is considered the actual measurement of LMA, then it may be useful to add a constant value of 1 cm² to each of the VIA measurements and to subtract a constant value of 4 cm² from each of the grid measurements to more closely approximate the defined actual value of LMA.

View larger version (26K): [in this window] [in a new window]   Figure 1. Frequency distribution of longissimus muscle area (LMA) values for video image analysis (VIA) instrument 1 and VIA instrument 2 subtracted from the corresponding average of three expert-gridded values or from the average of three planimeter readings of three acetate tracings per longissimus muscle by VIA procedure. Means for the two error groups differed (P < 0.05) for each procedure.

The mean absolute difference (mean absolute difference of each individual measurement from the average of the corresponding replicate measurements per LM and technique) was used as one method of determining how repeatable each of the procedural methods was in determining LMA (Table 5). Overall, mean absolute difference values among the various methods used to measure LMA were very low, with VIA instrument 2 in Procedure 3 having the highest value of 2.45 cm². The lowest VIA LMA mean absolute difference (difference of 0.39 to 0.58 cm² between image replicates) and standard deviation (0.65 to 0.90 cm²) was observed in Procedure 1. Procedure 1 may be adopted as a standard procedure to determine the repeatability of instrument-derived LMA measurements while carcasses are in a stationary position and would provide a baseline of the instrument’s repeatability without the error that might be introduced during operation. Video image analysis mean absolute difference values for Procedures 1 and 2 suggested that VIA instrumentation is highly repeatable when measuring LMA of stationary beef carcasses, outperforming the repeatability of expert-gridded and acetate/planimeter-traced measurements of LMA.

Variance components and the percentage of total variance accounted for by independent model variety for expert-gridded, acetate/planimeter-traced, and VIA Procedures 1, 2, and 3 LMA measurement methods are reported in Table 6. Across all LMA measurement methods, more than 89% of the total variance in LMA was accounted for by differences between carcass sides. The proportion of total variance due to VIA replicates for Procedures 1 and 2 (1.58 and 1.75%, respectively) was much smaller than the proportion of total variance due to measurement replicates by experts using a grid or the tracing method (2.84 and 5.30%, respectively). This indicated that less variance was associated with VIA measurement of LMA (in a stationary setting) than when compared with commonly used techniques for measuring LMA. Clearly, more variance was introduced in the VIA replicates when LMA was obtained at operational chain speeds (6.24%), further indicating the importance of standardized operator training.

	Implications

Top Abstract Introduction Materials and Methods Results and Discussion Implications Literature Cited

 
Video image analysis instrumentation can be used to assess beef carcass longissimus muscle area in both stationary and operational scenarios with high levels of accuracy and repeatability. Data from this study provide stationary and operational performance standards for the accuracy and repeatability of instrument-derived longissimus muscle area measurements. Instrument assessment of longissimus muscle area has the potential to improve the predictive ability of USDA yield grades in a yield grade augmentation system.

	Footnotes

 
¹ The authors wish to thank Excel Corporation for providing use of their facilities during data collection.

² Present address: Excel Corp., Wichita, KS 67219-7550.

³ Correspondence: 7C Animal Science Bldg. (phone: 970-491-5826; fax: 970-491-0278; E-mail: Keith.Belk{at}colostate.edu).

Received for publication December 6, 2002. Accepted for publication April 14, 2003.

	Literature Cited

Top Abstract Introduction Materials and Methods Results and Discussion Implications Literature Cited

 


Belk, K. E., J. A. Scanga, J. D. Tatum, J. W. Wise, and G. C. Smith. 1998. Simulated instrument augmentation of USDA yield grade application to beef carcasses. J. Anim. Sci. 76:522–527.[Abstract/Free Full Text]

Bodwell, C. E., G. Harrington, and R. W. Pomeroy. 1959. A note on the measurement of ‘eye’ muscle area in beef carcasses. Anim. Prod. 1:97–101.

Borggaard, C., N. T. Madsen, and H. H. Thodberg. 1996. In-line image analysis in the harvest industry, illustrated by beef carcass classification. Meat Sci. 43:S151–S163.

Butler, O. D., M. J. Garber, and R. L. Smith. 1956. Beef carcass composition and yield of wholesale cuts as estimated from left and right sides. J. Anim. Sci. 15:891–895.[Abstract/Free Full Text]

Cannell, R. C., J. D. Tatum, K. E. Belk, J. W. Wise, R. P. Clayton, and G. C. Smith. 1999. Dual-Component Video Image Analysis System (VIAscan^TM) as a predictor of beef carcass red meat yield percentage and for augmenting application of USDA Yield Grades. J. Anim. Sci. 77:2942–2950.[Abstract/Free Full Text]

Carpenter, J. W., and A. Z. Palmer. 1961. Variability of ribeye area and degree of marbling of beef carcasses as influenced by method of ribbing. J. Anim. Sci. 20:915. (Abstr.)

Cole, J. W., L. E. Orme, and C. M. Kincaid. 1960. Relationship of loin eye area, separable lean of various beef cuts and carcass measurements to total carcass lean in beef. J. Anim. Sci. 19:89–100.[Abstract/Free Full Text]

Cross, H. R., G. C. Smith, C. E. Murphey, D. M. Stiffler, L. W. Douglas, and J. W. Savell. 1984. USDA beef grades: An evaluation of the accuracy and uniformity of their application. J. Food Qual. 7:107–120.

Gardner, D. B., H. G. Dolezal, and D. M. Allen. 1995. Utilization of video image analysis (VIA) for predicting beef subprimal yields. J. Anim. Sci. 73(suppl. 1):164. (Abstr.)

George, M. H., H. G. Dolezal, J. D. Tatum, J. B. Morgan, J. W. Wise, C. R. Calkins, J. O. Reagan, and G. C. Smith. 1996. USDA yield grades, total body electrical conductivity and video image analysis technologies for predicting cutability of sides of steer/heifer carcasses. Pages 121–129 in Colorado State Univ. 1998 Beef Program Report. Dept. of Anim. Sci., Colorado State Univ., Ft. Collins.

Hedrick, H. B., J. C. Miller, G. B. Thompson, and R. R. Freitag. 1965. Factors affecting longissimus dorsi area and fat thickness of beef and relation between these measurements and retail yield. J. Anim. Sci. 24:333–337.[Abstract/Free Full Text]

Henderson, D. W., D. E. Goll, M. H. Stromer, M. J. Walter, E. A. Kline, and R. E. Rust. 1966. Effects of different measurement techniques and operators on bovine longissimus dorsi area. J. Anim. Sci. 25:334–338.[Abstract/Free Full Text]

McKenna, D. R., D. L. Roeber, P. K. Bates, T. B. Schmidt, D. S. Hale, D. B. Griffin, J. W. Savell, J. C. Brooks, J. B. Morgan, T. H. Montgomery, K. E. Belk, and G. C. Smith. 2002. National Beef Quality Audit—2000: Survey of targeted cattle and carcass characteristics related to quality, quantity, and value of fed steers and heifers. J. Anim. Sci. 80:1212–1222.[Abstract/Free Full Text]

Montgomery, D. C. 1997. Design and Analysis of Experiments. 4th ed. John Wiley & Sons, New York.

Naumann, H. D. 1951. A recommended procedure for measuring and grading beef for carcass evaluation. Proc. Recip. Meat Conf. 4:89–93.

This article has been cited by other articles:

				B. C. N. Cunha, K. E. Belk, J. A. Scanga, S. B. LeValley, J. D. Tatum, and G. C. Smith Development and validation of equations utilizing lamb vision system output to predict lamb carcass fabrication yields J Anim Sci, July 1, 2004; 82(7): 2069 - 2076. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Steiner, R. Articles by Smith, G. C. Search for Related Content PubMed PubMed Citation Articles by Steiner, R. Articles by Smith, G. C.