Fifty-one Churra da Terra Quente ewes (4–7 years old) were used to analyse the potential of real-time ultrasound (RTU) to predict the amount of internal adipose depots, in addition to carcass fat (CF). The prediction models were developed from live weight (LW) and RTU measurements taken at eight different locations. After correlation and multiple linear regression analysis, the prediction models were evaluated by k-fold cross-validation and through the ratio of prediction to deviation (RPD). All prediction models included at least one RTU measurement as an independent variable. Prediction models for the absolute weight of the different adipose depots showed higher accuracy than prediction models for fat content per kg of LW. The former showed to be very good or excellent (2.4 ⩽ RPD ⩽ 3.8) for all adipose depots except mesenteric fat (MesF) and thoracic fat, with the model for MesF still providing useful information (RPD = 1.8). Prediction models for fat content per kg of LW were also very good or excellent for subcutaneous fat, intermuscular fat, CF and body fat (2.6 ⩽ RPD ⩽ 3.2), while the best prediction models for omental fat, kidney knob, channel fat and internal fat still provided useful information. Despite some loss in the accuracy of the estimates obtained, there was a similar pattern in terms of RPD for models developed from LW and RTU measurements taken just at the level of the 11th thoracic vertebra. In vivo RTU measurements showed the potential to monitor changes in ewe internal fat reserves as well as in CF.

中文翻译：

---

使用实时超声对奶羊的胴体和内部脂肪库进行体内评估

51 只 Churra da Terra Quente 母羊（4-7 岁）被用于分析实时超声（RTU）预测除胴体脂肪（CF）之外的内部脂肪库数量的潜力。预测模型是根据在八个不同位置进行的活重 (LW) 和 RTU 测量值开发的。经过相关性和多元线性回归分析，预测模型通过以下方式评估：ķ- 折叠交叉验证和通过预测与偏差的比率 (RPD)。所有预测模型都包括至少一个 RTU 测量作为自变量。不同脂肪库绝对重量的预测模型显示出比每公斤 LW 脂肪含量的预测模型更高的准确性。对于除肠系膜脂肪 (MesF) 和胸部脂肪以外的所有脂肪库，前者显示非常好或非常好 (2.4 ⩽ RPD ⩽ 3.8)，MesF 模型仍然提供有用的信息 (RPD = 1.8)。每公斤 LW 脂肪含量的预测模型对于皮下脂肪、肌间脂肪、CF 和体脂（2.6 ⩽ RPD ⩽ 3.2）也非常好或优秀，而网膜脂肪、肾结、通道脂肪和内脏脂肪的最佳预测模型fat 仍然提供了有用的信息。体内RTU 测量显示了监测母羊体内脂肪储备和 CF 变化的潜力。