Ultrasound prediction of intramuscular fat content in lean cattle

Accuracy and precision of intramuscular fat prediction by ultrasound (USIMF) in lean cattle were studied by developing a prediction model based on 172 pure and crossbred beef and dual purpose bulls and steers from commercial herds. The animals were scanned across the 13th thoracic vertebrae using a Pie 200 SLC scanner. Five images were collected per individual. A 2.5 cm cross sectional muscle sample at the 13th thoracic vertebrae was collected from each carcass for chemical analysis of intramuscular fat % (CHIMF; Mean 2.0%, SD 1.5%) at the scanning site. The prediction model was developed using stepwise regression procedures. The validation model R(2), RMSE. overall mean bias, SEP and r(p) correlation between CHIMF and USIMF based on the final prediction model was 0.80, 0.66%, -0.15%, 0.65% and 0.90, respectively. The rank correlations varied over the CHIMF parameter space (from 0.26 to 0.87). The causes of this and the implications that follow are discussed. The overall results imply that ultrasound can predict IMF in lean cattle with acceptably high accuracy and precision. Ultrasound may thus be a valuable selection tool for meat quality in such cattle populations. The importance of developing optimised prediction models, based on data representative for the actual population, is highly emphasised. (C) 2009 Elsevier B.V. All rights reserved.