摘要
用直接注射的方式,将猪背最长肌制备成注水比例分别为0、2%、6%、10%和14%的肉样本。利用低场核磁共振(field nuclear magnetic resonance,LF-NMR)技术测量肉样本产生的NMR信号,并将反演后得到的横向弛豫谱参数作为自变量,通过判别分析(Discriminant analysis,DA)和偏最小二乘回归(partial least square regression,PLSR)分析,分别建立注水肉检测模型,并尝试多种可能性的PLSR建模,评估分析模型对较低注水比例的注水肉的检测性能。结果表明,对DA模型进行回代验证和留一交叉验证,模型对注水肉分类的总正确率分别为89.4%和88.2%。采用将杠杆值和学生化残差相结合的方式判别和删除异常数据,用变量投影重要性分析法筛选出6个横向弛豫谱参数作为自变量建立的优化PLSR模型的检测性能最优,校准集决定系数(R_(c)^(2))和标准误差(SEV)分别为0.9603和1.0033%,交叉验证的决定系数(R_(cv)^(2))和标准误差(SECV)分别为0.9508和1.1169%,预测集决定系数(R_(p)^(2))和标准误差(SEP)分别为0.9518和1.1280%。在95%的置信概率下,优化PLSR模型能够预测未知样本中注水百分比的置信区间的最好估计值约为2.256%。表明基于LF-NMR横向弛豫谱建立的DA模型和PLSR模型可以对注水肉进行有效的定性和定量检测。
The longissimus dorsi in pigs was prepared into the meat samples with water injection ratio of 0%,2%,6%,10%and 14%by direct injection.The NMR signals generated by the meat samples were measured by LF-NMR technique.The transverse relaxation spectrum parameters obtained by inversion were taken as independent variables,combined with discriminant analysis and partial least square regression(PLSR)analysis,the models for detecting water-injected meat were developed,and multiple possibilities for PLSR modeling were attempted.It was showed that the developed discriminant model was validated with calibration set and leave-one-out cross,the total accuracy of the classification of water-injected meat was 89.4%and 88.2%,respectively.By combining the lever value with the student residual,the abnormal data was distinguished and deleted,and the prediction performance of the PLSR model established by using the variable projection importance analysis method to select out six transverse relaxation spectrum parameters as independent variables was optimal,the determination coefficient(R_(c)^(2))and standard error(SEV)from calibration set were 0.9603 and 1.0033%,the determination coefficient(R_(cv)^(2))and standard error(SECV)from cross validation were 0.9508 and 1.1169%,the determination coefficient(R_(p)^(2))and standard error(SEP)from prediction set were 0.9518 and 1.1280%,respectively.The best estimate of the confidence interval capable of predicting the percentage of water injection in unknown samples was about 2.256%at 95%confidence probability.The results showed that the DA model and PLSR model based on LF-NMR transverse relaxation spectrum could be used for qualitative and quantitative detection of water-injected meat.
作者
许建梅
杨芳
林卿
王佳慧
艾竹君
XU Jianmei;YANG Fang;LIN Qing;WANG Jiahui;AI Zhujun(Modern Education Technology Center,Hainan Medical University,Haikou 571199,China;School of Biomedical Information and Engineering,Hainan Medical University,Haikou 571199,China;School of Tropical&Laboratory,Hainan Medical University,Haikou 571199,China)
出处
《食品工业科技》
CAS
北大核心
2021年第11期226-232,共7页
Science and Technology of Food Industry
基金
海南省自然科学基金项目(818MS065)
海南医学院教育科研课题(HYYB202032)。
关键词
低场核磁共振
判别分析
偏最小二乘回归
注水肉
low-field nuclear magnetic resonance
discriminant analysis
partial least square regression
water-injected meat
