Elsevier

Food Chemistry

Volume 293, 30 September 2019, Pages 204-212
Food Chemistry

Identification of antibiotic mycelia residues in cottonseed meal using Fourier transform near-infrared microspectroscopic imaging

https://doi.org/10.1016/j.foodchem.2019.04.100Get rights and content

Highlights

  • A method for identifying antibiotic mycelia residues adulterated in cottonseed meal is proposed.

  • The method is based on one-class partial least squares and near-infrared microspectroscopic imaging.

  • The characteristic bands are selected by principal component analysis.

  • The locations of abnormal spectra are displayed and visualized on the image of sample.

Abstract

Near-infrared microscopy (NIRM) technology can analyze different components within a sample while also obtaining spatial information about the sample. No rapid detection methods are available for effectively identifying antibiotic mycelia residues (AMRs) in protein feeds materials to date. In this study, the feasibility of using NIRM to identify AMRs (oxytetracycline residue, streptomycin sulfate residue and clay colysin sulfate residue) mixed in cottonseed meals was studied. The samples were scanned by NIRM, then the spectra of images were analyzed by principal component analysis (PCA) to select characteristic bands for further identification with one-class partial least squares analysis (OCPLS). The results showed that: a) AMRs were effectively identified in cottonseed meal; b) screening characteristic bands and increasing the spectral number of the calibration set improved the identification results of the model; and c) the sensitivity, specificity, accuracy and class error of the method were 100%, 95.93%, 99.01% and 2.03%, respectively.

Introduction

Antibiotic mycelia residues (AMRs) are the by-products of microbial fermentation for the production of antibiotics that are left after the separation of antibiotics (Shi, Ai, Wang & Sun, 2015). The main components of AMRs are microbial mycelium, unused medium residue, process metabolites and residual antibiotics. The crude protein content of AMRs is relatively high (Zhou, Gao, Wang, & Wang, 2011), about 30%–50% (Lin et al., 2018), and is similar to that of commonly used protein feeds (Sanchez et al., 2010). Because AMRs contain trace amounts of antibiotics (about 0.6 wt%) and have not been adequately evaluated for safety, there is a risk of inducing resistant bacteria. If AMRs are illegally used as feeds material, the feeds could pose some safety risks. In the process of microbial fermentation into antibiotics, other unknown intermediates are produced, and adding them into feeds before their safety is determined could seriously threaten the safety of feeds quality. The long-term use of animal feeds containing illegal AMRs would potentially allow any residual antibiotics in the AMRs to be transferred to meat, eggs, milk and other livestock and poultry products, which will also cause bacteria in the animal to become resistant because of prolonged contact with antibiotics (Yang et al., 2017). For those reasons, all types of AMRs have been listed in the National Hazardous Waste List of China since 2008, marked as “disposal only in landfillor by incineration”. Therefore, developing an effective method to identify AMRs in animal feeds is of great significance for ensuring the safety of feeds, aquaculture and food.

The composition of AMRs is very similar to protein feeds materials, and therefore it is difficult to distinguish between them. At present, the methods available to determine whether AMRs have been added to the feeds are by detecting the presence of antibiotics or antibiotic markers (Yuan, 2014, Li, 2017). However, these methods have many limitations: they cannot determine the source (from AMRs or other components) of antibiotics, the types of antibiotics in the feeds needs to be known in advance before the sample is tested, and the methods are cumbersome, time-consuming and expensive. The current lack of methods for rapid detection of AMRs in feeds highlights the need to develop a convenient, efficient and non-destructive method for accurately determining AMRs in feeds.

Near-infrared spectroscopy (NIRS) technology is a green, efficient, and non-destructive analytical method widely used for quality control and adulteration detection (Haughey et al., 2013, Wang et al., 2018). Near-infrared microscopy (NIRM) is an effective combination of NIR and optical microscopy. When compared with other qualitative detection methods, NIRM provides spatial and visual information by simple manipulation and does not require sample pretreatment, making it advantageous especially for in situ analyses (Ravikanth et al., 2017, Riedl et al., 2015). The data from NIRM are three-dimensional (x × y × z), with x  × y representing the spatial information of the sample, and z representing the spectral information of the sample, with each spectrum representing the sample information within each pixel. This provides the position information of the abnormal component, which can be converted into a visual image (Munera et al., 2018, Zhang et al., 2017). Because of these advantages, NIRM technology has been widely used in cereals, food, feed, medicine and other fields (Fu et al., 2014, Barbin et al., 2012, Baca-Bocanegra et al., 2018).

Cottonseed meal, consisting of black cottonseed hulls and deep yellow cottonseed kernels, is commonly used as a protein feeds ingredient. Cottonseed hulls in cottonseed meals look very similar to AMRs from their appearance, both being black or dark brown particles. The adulteration of cottonseed meal by AMRs is difficult to identify by direct observation with the eye. Because cottonseed kernels are dark yellow, they can be visually distinguished from cottonseed hulls and AMRs by their appearance. Therefore, when determining if cottonseed meal has been adulterated with AMRs, it is only necessary to analyze the black or black-brown particles using NIRM.

The present study used a Fourier transform near-infrared microscopy imaging system to collect visible images and NIRM images of cottonseed hulls, cottonseed kernels, AMRs, and adulterated samples. The spectra of the region of interest (ROI) in the NIRM images was extracted. Qualitative discriminant analysis models of cottonseed hulls and different AMRs were established using characteristic bands selected by Principal Component Analysis (PCA) combined with One-class partial least squares analysis (OCPLS), and the feasibility of discriminating ARMs mixed in adulterated samples was explored.

Section snippets

Sample collection and preparation

The three types of AMR used in the experiment were collected from different antibiotic manufacturers in Henan and Fujian provinces, China, namely oxytetracycline residue (OR), streptomycin sulfate residue (SR), and clay colysin sulfate residue (CR); all samples were collected in the wet state and were then air dried. Seven samples of cottonseed meal (I–VII), were collected from Shandong and Henan provinces; and six cottonseed meals (I–VI) were used to establish the OCPLS model; cottonseed meal

Spectral analysis of cottonseed meals and AMRs

The cottonseed kernels are dark yellow, the OR, the SR and the cottonseed hulls are black, and the CR are dark brown (Fig. 1a). The cottonseed kernels differ significantly from the other four samples in color, so when the AMRs are adulterated into the cottonseed meals, the cottonseed kernels can be visually distinguished, but cottonseed hulls and AMRs cannot be distinguished; therefore, other technical methods are needed for discrimination.

The mean and the second derivative spectra of

Conclusions

This study investigated the feasibility of effectively distinguishing AMRs (oxytetracycline residue, streptomycin sulfate residue and clay colysin sulfate residue) mixed in cottonseed meals using NIRM imaging technology. In this experiment, all samples were ground to ensure the sample particles on the background plate were 1–2 mm, then the NIRM images of the sample particles were obtained. By using PCA and OCPLS, different types of AMRs could be effectively distinguished in the adulterated

Acknowledgments

The authors gratefully acknowledge financial support from the National Key R&D Program of China (2017YFE0115400) and Program for Changjiang Scholars and Innovative Research Team in University of Ministry of Education of China (IRT-17R105). We thank Philip Creed, PhD, from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.

Conflict of interest

The authors have declared no conflict of interest.

References (28)

  • I. Stanimirova et al.

    A comparison between two robust PCA algorithms

    Chemometrics and Intelligent Laboratory Systems

    (2004)
  • L. Xu et al.

    One-class partial least squares (OCPLS) classifier

    Chemometrics and Intelligent Laboratory Systems

    (2013)
  • M. Zhang et al.

    Classification of foreign matter embedded inside cotton lint using short wave infrared (SWIR) hyperspectral transmittance imaging

    Computers and Electronics in Agriculture

    (2017)
  • Diao Qi et al.

    The effect of apple’s fermentation on performance and immunity for lactating dairy cattle

    China Dairy Cattle

    (2003)
  • Cited by (3)

    • Machine Learning Based 3D in Situ Visual Discriminant Analysis of Mammalian and Non-mammalian Bone Meals by Micro-CT

      2023, Nongye Jixie Xuebao/Transactions of the Chinese Society for Agricultural Machinery
    1

    These authors contributed equally to this work.

    View full text