Scholarlink

<p>This study establishes a method for tracing the origin of Pacific oysters based on transmission mid-infrared spectroscopy (TMIRS) combined with multivariate statistical techniques. The acquired mid-infrared (MIR) spectral data were preprocessed by absorbance transformation (AbT), automatic baseline correction (AutoBC), etc. Then, principal component analysis (PCA) was performed in the wavenumber ranges of 3800 –2300&#xa0;cm<sup>−1</sup> and 1900 − 800&#xa0;cm<sup>−1</sup>. Taking the principal component scores and their combinations as input variables, four modeling methods including Fisher Discriminant Analysis (FDA), Partial Least Squares Discriminant Analysis (PLS-DA) Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) and BP Neural Network (BPNN) were used to conduct the origin tracing research. The results indicate that AbT is the optimal preprocessing method. The accuracy rates of the training sets for all four models are exceeding 98%, and the accuracy rates of the test sets are higher than 97%, all four models are applicable to the origin traceability of pacific oysters. The combination of TMIRS and multivariate statistical methods provides a convenient and efficient tracing technique, offering technical support for safeguarding consumers’ rights and interests and promoting the sustainable development of the oyster industry.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Transmission Mid-Infrared Spectroscopy Combined with Multivariate Statistical Methods to Trace the Origin of Pacific Oysters from Different Regions in China

  • 期刊论文

  • 发表日期：

    2025年11月4日

查看全文
  • Wanli Zong,
  • Zilong Zhu,
  • Yi Xu,
  • Jinyu Mu,
  • Xinpu Guo,
  • Gang Lu,
  • Yezhong Xing

摘要

This study establishes a method for tracing the origin of Pacific oysters based on transmission mid-infrared spectroscopy (TMIRS) combined with multivariate statistical techniques. The acquired mid-infrared (MIR) spectral data were preprocessed by absorbance transformation (AbT), automatic baseline correction (AutoBC), etc. Then, principal component analysis (PCA) was performed in the wavenumber ranges of 3800 –2300 cm−1 and 1900 − 800 cm−1. Taking the principal component scores and their combinations as input variables, four modeling methods including Fisher Discriminant Analysis (FDA), Partial Least Squares Discriminant Analysis (PLS-DA) Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) and BP Neural Network (BPNN) were used to conduct the origin tracing research. The results indicate that AbT is the optimal preprocessing method. The accuracy rates of the training sets for all four models are exceeding 98%, and the accuracy rates of the test sets are higher than 97%, all four models are applicable to the origin traceability of pacific oysters. The combination of TMIRS and multivariate statistical methods provides a convenient and efficient tracing technique, offering technical support for safeguarding consumers’ rights and interests and promoting the sustainable development of the oyster industry.