Evaluation of three ultrasound instruments, CVT-2, UltraFom 300 and AutoFom for predicting lean in the primals salable meat yield and weight of of pork carcasses

Three commercially available ultrasound instruments were evaluated using pigs representative of the Canadian pig population: CVT-2, UltraFom 300 and AutoFom. The reflectance probe HGP2 was utilized as the baseline instrument. Sampling was stratified by fat thickness. The original data set (n = 236) was divided randomly into a calibration set (n = 194 [85 barrows and 109 gilts]) to calibrate the instruments and a validation set (n = 72 [32 barrows and 40 gilts]) for validating the calibration models. For salable meat yield, RMSE values for the calibration models were: HGP2, 1.56; CVT-2, 1.57; UltraFom, 1.70; and AutoFom, 1.68. For lean (kg) in butt, picnic, loin and ham, and weight (kg) of the skinless, trimmed belly, RMSE for the calibration models was similar for all four instruments: respectively, 0.19-0.21, 0.21-0.23, 0.31-0.37, 0.35-0.40 and 0.35-0.36 kg. Validation results for predicting salable meat yield show that the improvement in precision and/or accuracy associated with the UltraFom 300 or AutoFom, over the baseline reflectance probe HGP2, was negligible whereas a definite advantage was observed for the CVT-2. However, validation results for predicting kg of lean in the primals did not show any clear advantage for the three commerciallly available ultrasound instruments over the baseline reflectance probe HGP2. Calibration and validation procedures showed that, if one was to base the assessment of these three ultrsound instruments uniquely on the precision and accuracy for predicting salable meat yield and/or kg of lean in the primals, any amelioration over the baseline reflectance probe HGP2 would have to be viewed as rather negligible; particularly for the UltraFom 300 or AutoFom. (C) 2004 Elsevier Ltd. All rights reserved.