基于低场核磁共振技术的小鼠体成分无损分析方法开发

Development of a Noninvasive Analytical Method of Body Composition in Mice Based on Low Field Nuclear Magnetic Resonance Technique

本研究利用传统的全身化学分析方法,包括干燥法、索氏提取法、凯氏定氮法、灼烧法测定的小鼠样品的体成分含量作为参考值,结合低场核磁共振技术及偏最小二乘回归法(PLSR)和主成分回归法(PCR),分别建立了样品体液、脂肪、瘦肉含量的预测模型。结果显示,PLSR预测模型的校正集相关系数(R_(cal)~2)和交互验证集相关系数(R_(cv)~2)均大于0.98,且标准误差值较低,说明该模型的稳定性较好,通过该模型可以准确预测小鼠体液、脂肪和瘦肉含量。对未参与建模的小鼠进行体液、脂肪和瘦肉含量的测定,其相对误差均小于1.5%,表现出较好的应用潜力。

It is of great importance for food component evaluation to develop a fast and noninvasive method for prediction of body composition content（body fluid,body fat and lean）in mice.The body composition in unknown mice was predicted by the developed model.The body composition determined by the traditional whole-body chemical analysis methods including drying,Soxhlet extraction method,Kjeldahl Test and burning was used as reference,and the calibration and validation models of body fluid,fat and lean mass were developed by the NMR combined with the partial least square regression（PLSR）and principal component regression（PCR）,respectively.The determination coefficients R^2 of calibration models and validation models were greater than 0.98 with a relatively low standard deviation,indicating good stability of the developed model.Moreover,the models have a good prediction ability of body composition in mice.The prediction of body composition for unknown mice showed a relative standard deviation less than 1.5% as compared with the traditional method,revealing their good potential in analysis of body composition in mice.