期刊
MICROCHEMICAL JOURNAL
卷 149, 期 -, 页码 -出版社
ELSEVIER
DOI: 10.1016/j.microc.2019.104057
关键词
Sorghum bicolour (L.) Moench; Infested grains; Food pest; NIR; Chemometrics; Supervised classification
The potential of near-infrared spectroscopy (NIRS) combined with partial least squares discriminant analysis (PIS-DA) to develop a screening method for distinguishing uninfested from infested sorghum (Sorghum bicolour (L.) Moench) grains was for the first time investigated. A total of 108 sorghum grain samples from thirty-six different genotypes were infested with seventy Sitophilus zeamais Motschulsky (Coleoptera: Curculionidae) unsexed adult insects during seventy days. More 101 uninfested sorghum grain samples from these same genotypes were used to build models. Principal components analysis (PCA) allowed slightly discriminating between the two classes along principal component two. A PIS-DA model presented perfect classification rates, with sensitivity and specificity equal to 100% for the test set. In addition, the model showed high accuracy, and accordance and concordance (precision) both equal to 100%. These results showed that the combination of NIBS with PLS-DA provided a rapid, cost-effective and non-invasive way to detect insect infestation in sorghum grains.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据