Workpackage 3

Paul Allen

New approaches to determine the reference



To develop indirect methods of predicting the lean meat percentage of a pig carcass that are highly correlated with the reference dissection method. These methods will be lower cost than the existing dissection method and could potentially be more repeatable and reproducible. Following the successful completion of this task an assessment of the relative merits of the different methods and the implications of their adoption for authorizing and monitoring classification equipment will be made.


The calibration of pig carcass grading instruments involves the dissection of at least 120 half carcasses. This is costly and time consuming and requires highly skilled labour to achieve consistent results. If an instrumental method of measuring the lean meat content of a carcass were available that was at least as accurate as the EU reference dissection method then more consistent grading could be expected since the human error in determining the reference would be removed. Also, if such a method were less costly then a larger sample size could be used for the calibration and the validity of the regression equation could be checked more frequently. Furthermore, the availability of an instrumental method of determining the lean meat percentage rather than the reference dissection method could make it easier for the applicant countries to comply with the grading regulations.


Three indirect methods that have the potential to predict lean meat percentage in a carcass with a high degree of accuracy are being evaluated. These are a vision system, X-ray CT and MRI. The vision system has been developed by Cemagref to estimate the lean meat percentage of cuts from images of cut faces. This will be adapted to take images of a larger number of cut faces in order to find a combination of images that can accurately predict the lean meat percentage in the whole carcass. The University of Kaposvar has both X-ray CT and MRI facilities available for use in the project at its diagnostic centre. These are non-invasive imaging methods with the potential to take a large number of cross sectional images through the carcass. The relatively long acquisition and relaxation times for MRI limits the number of images that can be taken. X-ray CT is much faster and so contiguous slices through the whole carcass will be taken. From these in addition to the muscle and fat areas in individual slices the total volumes of muscle and fat in the entire carcass will be calculated.

A total of 120 carcasses in two equal batches will be used. These will be from three distinct genotypes (Dalland and Seghers hybrids and Large White/Landrace crosses). These will be selected at the Kometa abattoir in Kaposvar. There will be approximately equal numbers of females and castrated males. Care will be taken to select carcasses that are evenly split. The left side of each carcass will be dissected according to the reference method (four main cuts) by the team of 8 butchers used in Workpackage 1 to obtain the reference lean meat percentage. The remainder of the cuts will also be dissected to obtain the total lean meat percentage in the carcass. The right side of each carcass will be assessed by the three indirect methods. Variables from each indirect method will be used to predict the EU reference lean percentage and the total lean percentage. Recommendations on the best method to use in place of dissection in future calibration trials will be based on the accuracy of prediction of lean meat percentage and the cost per carcass.