Prediction of pork carcass composition based on cross-sectional region analysis of dual energy X-ray absorptiometry (DXA) scans § A.D. Mitchell a, *, A.M. Scholz b , V.G. Pursel a a Growth Biology Laboratory, Agricultural Research Service, USDA, Bldg. 200, Rm. 205, BARC-East, Beltsville, MD 20705, USA b Ludwig-Maximilians University-Munich, 85764 Oberschleissheim, Germany Received 3 January 2002; received in revised form 15 March 2002; accepted 15 March 2002 Abstract Dual energy X-ray absorptiometry (DXA) was used to measure pork carcass composition by performing a total scan of the right half of 262 pork carcasses (42.7 Æ 5.2 kg). The DXA scans were analyzed for percentage fat in the entire half-carcass as well as the shoulder, ham, loin, and side regions. In addition, a total of 14 cross-sections (57.6 mm wide) were analyzed: six in the shoulder/ thoracic region, three in the loin region, and five in the ham region. Relative to the DXA measurement of total fat content, the coefficient of determination (R 2 ) for a single cross-sectional slice ranged from 0.908 to 0.976. Relative to chemical analysis, a single slice from the ham region predicted the percentage of fat or lean in the half-carcass with an R 2 of 0.81 and a standard error of the estimate of 2.04. Prediction equations were used to analyze a separate group of 65 half-carcasses. These results indicate that carcass fat and lean percentages can be measured by performing a single-pass cross-sectional scan that would be compatible with on-line processing. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Carcass Composition; Dual energy X-ray absorptiometry; Pigs 1. Introduction A rapid, accurate method is needed to provide infor- mation regarding the fat and lean content during the on- lineprocessingofporkcarcasses.Themethodshouldalso be non-invasive and require little user input in terms of manipulationanddataprocessing.Amongthetechniques that have been evaluated recently are: total body elec- tricalconductivity(TOBEC;Berg,Engel&Forrest,1998; Berg, Forrest, & Fisher, 1994); bioelectrical impedance (BIA; Swantek, Crenshaw, Marchello, & Lukaski, 1992; Swantek, Marchello, Tilton, & Crenshaw, 1999); ultra- sound (Brøndum, Egebo, Agerskov, & Busk, 1998; Liu & Stauffer, 1995); and video image analysis (VIA; Bransc- heid & Dobrowolski, 1996; Branscheid, Dobrowolski, & Ho¨reth, 1995; Stanford, Richmond, Jones, Robert- son, Price, & Gordon, 1998). Other techniques may be sufficiently accurate but have limitations that preclude serious consideration for on-line adaptation (Forrest, Kuei, Orcutt, Schinckel, Stouffer, & Judge, 1989). Mainly, two techniques have been evaluated for automated on-line measurement of pork carcass com- position. Using electromagnetic scanning or TOBEC in an on-line industrial configuration, Berg et al. (1994) reported the correlation (r) of a variety of TOBEC readings with carcass measurements of total dissectible carcass lean (0.54–0.93), percentage of carcass lean (0.29–0.75), and total fat (À0.09–0.66). The same study reported prediction equations for total dissected carcass lean (R 2 =0.83–0.90) and percentage of carcass lean (R 2 =0.82–0.86). Ultrasound technology has been suc- cessfullyadaptedforon-lineporkcarcassevaluation.Liu and Stouffer (1995), using an ultrasound with a 3.5-MHz linear array transducer, reported an R 2 of 0.88 for pre- diction of carcass lean weight. A fully automated system consisting of 16–2.0 MHz transducers described by Brondum et al. (1998) predicted the meat percentage of pork carcasses with an accuracy (RSD) of 1.58–1.95%. Dual energy X-ray absorptiometry (DXA) can be used to measure pork carcass composition by perform- ing a total scan of the half-carcass (Mitchell, Scholz, 0309-1740/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0309-1740(02)00081-5 Meat Science 63 (2003) 265–271 www.elsevier.com/locate/meatsci § Mention of a trade name does not constitute a guarantee or war- ranty by the USDA and does not imply its approval to the exclusion of other products that may be suitable. * Corresponding author. Tel.: +1-301-504-8868; fax: +1-301-504- 8623. E-mail address: mitchell@anri.barc.usda.gov (A.D. Mitchell).