Dual-energy X-ray absorptiometry: an effective approach for predicting broiler chicken body composition

Abstract

Two trials were carried out to develop and validate linear regression equations for body composition prediction using Dual-energy X-ray absorptiometry (DEXA). In Trial 1, 300 Cobb500 male chickens raised from 1 to 42 d of age were scanned in DEXA to estimate total weight, fat mass, soft lean tissue (SLT) mass, bone mineral content (BMC), and fat percentage. DEXA estimates were compared to body ash, crude fat, SLT (sum of protein and water) and scale body weight. The dataset was split, with 70% used for prediction equations development and 30% for testing, and the 5k-fold cross-validation analysis was used to optimize the equations. The R2, mean absolute error (MAE), and root-mean-squared error (RMSE) were used as precision and accuracy indicators. A negative correlation (ρ = -0.27) was observed for ash content, while no correlation was observed for protein content (P > 0.05). Predictive linear equations were developed to assess broiler weight (R2 = 0.999, MAE = 25.12, RMSE = 38.99), fat mass (R2 = 0.981, MAE = 13.87, RMSE = 21.28), ash mass (R2 = 0.956, MAE = 3.98, RMSE = 5.61), SLT mass (R2 = 0.997, MAE = 35.73, RMSE = 52.45), water mass (R2 = 0.997, MAE = 29.56, RMSE = 43.94), protein mass (R2 = 0.989, MAE = 12.94, RMSE = 19.05), fat content (R2 = 0.855, MAE = 0.81, RMSE = 1.05), SLT content (R2 = 0.658, MAE = 1.01, RMSE = 1.28), and water content (R2 = 0.678, MAE = 0.99, RMSE = 1.27). All equations passed the test. In Trial 2, 395 Cobb500 male chickens were raised from 1 to 42 d of age and used for validation of prediction equations. The equations developed for weight, fat mass, ash mass, SLT mass, water mass, and protein mass were validated. In conclusion, DEXA was found to be an effective approach for measuring the body composition of broilers when using predictive equations validated in this study for estimate calibration.