X-ray beef grading

Developed by researchers to quickly estimate 
carcass composition, it could some day usher in 
robotic cutting into beef processing plants

Oscar Lopez Campos explains the potential for dual energy X-ray absorptiometry (DXA).

Five years, 334 beef carcasses, 212 pork carcasses and 155 lamb car­cas­ses later, the meat science team at Agriculture and Agri-Food Can­ada’s Lacombe Research and Development Centre has proof that dual energy X-ray absorptiometry (DXA) technology is capable of accurately estimating carcass composition.

DXA units might best be described as oversized scanners. Their use in human medicine has been well-established for estimating body composition by emitting low- and high-energy wavelengths across the body and measuring the gradual reduction of the energy’s intensity from lean, fat and bone tissues. The rate of energy loss differs for each type of tissue making it possible to distinguish one from the other.

The research team’s job is to build prediction equations to assess the DXA unit’s potential for estimating the carcass composition of each species, explains Oscar Lopez Campos, the centre’s certified grader and carcass-merit researcher on this project.

This process requires massive amounts of data obtained by DXA scanning each carcass followed by manual dissection to separate and weigh lean, fat (subcutaneous, inter-muscular, body cavity) and bone content as the gold-standard cross-reference.

Since the typical horizontal DXA table designed for people isn’t large enough to accommodate a whole side of beef, the sides had to be fabricated into primals (chuck, rib, brisket, flank, foreshank, loin, round, plate). Each primal from the left side, which is the yield grading side at commercial plants, was scanned and dissected.

Samples for building the prediction equations also need to be representative of all types of carcasses graders would see in a commercial setting. To achieve this mix over the five years, 230 crossbred steers from the centre’s 300-head Angus-Simmental breeding herd were finished on a common ration at the onsite feedlot to weights ranging from 900 to 1,600 pounds with grade-fat thicknesses from two to 30 millimetres. The project included 104 cow carcasses as well and involved everyone starting with the beef production staff on through meat processing to all of the technicians who assist the scientists.

Equations developed at Lacombe can interpret DXA scans to reliably estimate lean and fat yield. photo: Debbie Furber

Results as of summer 2017, presented at the International Congress of Meat Science and Technology in Ireland, show that the equations developed during the study predicted lean and fat yield with high reliability (R2=0.99) relative to manual dissection for beef steers and cows. Findings so far also suggest that DXA’s capacity to estimate beef carcass composition is independent of maturity.

The results for pork (R2=0.99) and lamb (R2=0.97) fat and lean were equally as encouraging.

The equation for predicting bone content showed slightly lower accuracy than those for fat and lean estimations, but R2 values were still over 0.92 for steer and cow carcasses. This is most likely because the DXA scanner is currently calibrated for assessing bone mineral content and density in people, not for estimating whole bone content in livestock.

From research to reality

In summary, Lopez Campos says, so far the project has successfully developed robust equations for the use of DXA technology for estimating total lean and fat across the range of carcasses normally seen in the Canadian market.

The next step will be to validate the prediction accuracies in a commercial setting and to obtain calibration curves for specific retail cuts or carcass cut-out specifications.

Although the potential economic value is difficult to quantify at this early stage of DXA technology development for the beef industry, he says the equations have immediate use in research projects involving large numbers of animals, such as genomics and feeding studies, because DXA scans would eliminate much of the need for time-consuming, expensive and wasteful manual dissections or chemical analysis for carcass composition.

Additionally, routine use of non-destructive DXA technology on an ongoing basis would allow lean-yield algorithms to be improved and updated regularly as the genetic composition of cattle shifts or market needs change, thereby potentially replacing the need for expensive and sporadic national beef cut-outs. The Canadian yield equation built into the yield ruler, and more recently used to calibrate the e+v ribeye grade camera, was developed at AAFC-Lacombe based on a cut-out of 540 carcasses 25 years ago (1993).

Already at AAFC-Lacombe, DXA data is linked to both the ribeye and whole-carcass grade camera systems for continuing work with the Canadian industry on strengthening or developing prediction equations for the cameras.

Likewise, over the longer term DXA technology has potential to be linked to grade camera systems at commercial plants to regularly update yield algorithms, optimize carcass cut-outs and provide precise carcass yield estimations.

The team also sees possibilities down the road for the use of DXA images to mark anatomical positions for cutting to implement robotics in beef processing plants. Work along this line has advanced in Australia and New Zealand where Scott Technologies is developing an upright DXA scanner.

DXA technology was first evaluated in the late 1990s for analyzing body composition of poultry and pigs, and it has been more than a decade since researchers at AAFC-Lennoxville showed its potential for measuring pig and lamb carcass composition. Until the AAFC-Lacombe project, only a few studies around the globe had dealt with the use of DXA technology to analyze body composition of beef cattle. For the most part, those studies compared DXA values to chemical analysis of body composition rather than to carcass yield.

Lopez-Campos notices increasing interest in the use of DXA technology for predicting overall carcass composition. The main reasons are that the cost per unit scanned is becoming more affordable, the unit is easy to use compared to other technologies, such as CT (computed tomography) scans, and lots of data can be collected quickly without interrupting the flow of production. As with other electronic instruments, the assessment is objective and the images and resulting assessments can be stored for a permanent record and, if desired, easily shared online.

This article was first published on AGCanada.com.

About the author



Stories from our other publications