Lean content prediction in pig carcasses, loin and ham by computed tomography (CT) using a density model

Abstract

Pig carcasses (122 half carcasses, 52 hams and 52 loins) from the Spanish pig population, were obtained in a commercial slaughterhouse and scanned by computed tomography to generate a predictive model determining weight and lean content. The model is mainly based on a density correction equation. The weight prediction model used the area of the histogram of the whole half carcass in a range of -250 to +800 Hounsfield units added to 2769g corresponding to the average weight of the head and pig feet that have not been scanned. The lean content predictive model is based on the ratio between the area of the lean peak in the calculated histograms and the area of the histogram of the whole half carcass. Both models were correlated with a manual dissection of the samples. Results from the predictive models and from the dissection were compared with the calculation of the root mean square error of calibration (RMSEC) for weight determination and lean content. Results show that a RMSEC of 0.6kg can be obtained for the weight half carcass. For prediction of the lean meat percentage a RMSEC of 1.48% can be obtained for the carcasses, 0.97% for the ham and 1.07% for the loin. According to our results, with a simple methodology it is possible to have good prediction values of weight and lean percentage in accordance with EU regulation.