Murdoch University, School of Veterinary & Life Sciences, Western Australia 6150, Australia; Advanced Livestock Measurement Technologies Project, Murdoch, Australia.
Murdoch University, School of Veterinary & Life Sciences, Western Australia 6150, Australia; Advanced Livestock Measurement Technologies Project, Murdoch, Australia.
Meat Sci. 2021 Mar;173:108400. doi: 10.1016/j.meatsci.2020.108400. Epub 2020 Dec 3.
This experiment assessed the ability of an on-line dual energy x-ray absorptiometer (DEXA) installed at a commercial abattoir to determine commercial cut weights in lamb carcases at abattoir chain-speed. 200 lamb carcases were scanned using a DEXA that was trained to predict the computed tomography determined proportions of fat, lean, and bone. Models were then trained using hot carcase weight and, DEXA fat% value or GR tissue depth to predict cut weight. Results from validation tests of DEXA models demonstrated excellent precision for predicting cut weight, in most cases describing more than 85% of the variation, and RMSE values that represented between 5 and 13% of the average weight of each cut. For most cuts these weight predictions were superior to those informed by GR tissue depth. This precision was maintained upon validation. Additional analyses utilised pixel information from the fore, saddle, and hind sections of DEXA images. This further enhanced the predictive power of cut weight models.
本实验评估了安装在商业屠宰场的在线双能 X 射线吸收仪（DEXA）在屠宰场链速下确定羊肉胴体商业切割重量的能力。使用经过训练以预测计算机断层扫描确定的脂肪、瘦肉和骨骼比例的 DEXA 对 200 个羊胴体进行了扫描。然后使用热胴体重量和 DEXA 脂肪%值或 GR 组织深度训练模型，以预测切割重量。DEXA 模型验证测试的结果表明，预测切割重量的精度非常高，在大多数情况下，描述了 85%以上的变化，而 RMSE 值代表了每个切割平均重量的 5%至 13%。对于大多数切割，这些重量预测优于 GR 组织深度提供的预测。这种精度在验证过程中得以保持。其他分析利用了 DEXA 图像前、鞍和后部分的像素信息。这进一步提高了切割重量模型的预测能力。
