Rapid and accurate identification and quantification of Lycium barbarum L. components: Integrating deep learning and NMR for nutritional assessment.

Journal: Food research international (Ottawa, Ont.)
PMID:

Abstract

Lycium barbarum L. (L. barbarum), revered for its nutritional and commercial value, exhibits variable nutritional contents depending on the consumption method. This study introduces an innovative approach, the Identification and Quantification of L.barbarum Components (IQ-LC) model, for rapid and accurate identification and quantification analysis of L.barbarum components by integrating NMR spectroscopy with a multi-label one-dimensional convolutional neural network. This model demonstrated exceptional performance in identifying 25 known-concentration mixtures, achieving an accuracy of 99.74 %, a true positive rate of 97.89 %, a true negative rate of 99.94 %, a root mean squared error (RMSE) of 0.15, and a coefficient of determination (R) of 0.96. This method was then applied to analyze the nutritional content of L.barbarum across different consumption forms: fresh berries, puree, and tea. Fresh L.barbarum, though nutrient-rich, faces challenges related to transportation and storage. In contrast, L. barbarum tea exhibited the lowest nutrient levels. Therefore, L. barbarum puree is recommended as the most practical option for daily dietary supplementation. Finally, the IQ-LC model was employed to assess and compare the nutritional contents of L.barbarum puree from ten commercial brands available on the market, providing a fast and reliable method for both identification and quantification purposes of L.barbarum's nutritional components across various consumption methods. This study not only offers a novel tool for the market regulation of L.barbarum products but also contributes to the broader application of deep learning and NMR in the field of food science and nutritional analysis.

Authors

  • Chengcheng He
    Department of Electronic Science, Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen 361005, China.
  • Fengji Liu
    Department of Electronic Science, Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen 361005, China.
  • Xin Shi
    Department of Automation, Faculty of Information Technology, Beijing University of Technology, Beijing 100124, China.
  • Feng Xia
  • Liubin Feng
  • Guiping Shen
    Department of Electronic Science, Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen 361005, China.
  • Jianghua Feng
    Department of Electronic Science, Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen 361005, China. Electronic address: jianghua.feng@xmu.edu.cn.