Machine Learning Applied to the Detection of Mycotoxin in Food: A Systematic Review

Mycotoxins, toxic secondary metabolites produced by certain fungi, pose significant threats to global food safety and public health. These compounds can contaminate a variety of crops, leading to economic losses and health risks to both humans and animals. Traditional lab analysis methods for mycotoxin detection can be time-consuming and may not always be suitable for large-scale screenings. However, in recent years, machine learning (ML) methods have gained popularity for use in the detection of mycotoxins and in the food safety industry in general due to their accurate and timely predictions. We provide a systematic review on some of the recent ML applications for detecting/predicting the presence of mycotoxin on a variety of food ingredients, highlighting their advantages, challenges, and potential for future advancements. We address the need for reproducibility and transparency in ML research through open access to data and code. An observation from our findings is the frequent lack of detailed reporting on hyperparameters in many studies and a lack of open source code, which raises concerns about the reproducibility and optimisation of the ML models used. The findings reveal that while the majority of studies predominantly utilised neural networks for mycotoxin detection, there was a notable diversity in the types of neural network architectures employed, with convolutional neural networks being the most popular.

Prediction of Deoxynivalenol Contamination in Wheat via Infrared Attenuated Total Reflection Spectroscopy and Multivariate Data Analysis

The climate crisis further exacerbates the challenges for food production. For instance, the increasingly unpredictable growth of fungal species in the field can lead to an unprecedented high prevalence of several mycotoxins, including the most important toxic secondary metabolite produced by Fusarium spp., i.e., deoxynivalenol (DON). The presence of DON in crops may cause health problems in the population and livestock. Hence, there is a demand for advanced strategies facilitating the detection of DON contamination in cereal-based products. To address this need, we introduce infrared attenuated total reflection (IR-ATR) spectroscopy combined with advanced data modeling routines and optimized sample preparation protocols. In this study, we address the limited exploration of wheat commodities to date via IR-ATR spectroscopy. The focus of this study was optimizing the extraction protocol for wheat by testing various solvents aligned with a greener and more sustainable analytical approach. The employed chemometric method, i.e., sparse partial least-squares discriminant analysis, not only facilitated establishing robust classification models capable of discriminating between high vs low DON-contaminated samples adhering to the EU regulatory limit of 1250 μg/kg but also provided valuable insights into the relevant parameters shaping these models.

Combining AI Tools with Non-Destructive Technologies for Crop-Based Food Safety: A Comprehensive Review

On a global scale, food safety and security aspects entail consideration throughout the farm-to-fork continuum, considering food's supply chain. Generally, the agrifood system is a multiplex network of interconnected features and processes, with a hard predictive rate, where maintaining the food's safety is an indispensable element and is part of the Sustainable Development Goals (SDGs). It has led the scientific community to develop advanced applied analytical methods, such as machine learning (ML) and deep learning (DL) techniques applied for assessing foodborne diseases. The main objective of this paper is to contribute to the development of the consensus version of ongoing research about the application of Artificial Intelligence (AI) tools in the domain of food-crop safety from an analytical point of view. Writing a comprehensive review for a more specific topic can also be challenging, especially when searching within the literature. To our knowledge, this review is the first to address this issue. This work consisted of conducting a unique and exhaustive study of the literature, using our TriScope Keywords-based Synthesis methodology. All available literature related to our topic was investigated according to our criteria of inclusion and exclusion. The final count of data papers was subject to deep reading and analysis to extract the necessary information to answer our research questions. Although many studies have been conducted, limited attention has been paid to outlining the applications of AI tools combined with analytical strategies for crop-based food safety specifically.

Low-Cost Pocket Fluorometer and Chemometric Tools for Green and Rapid Screening of Deoxynivalenol in Durum Wheat Bran

Cereal crops are frequently contaminated by deoxynivalenol (DON), a harmful type of mycotoxin produced by several Fusarium species fungi. The early detection of mycotoxin contamination is crucial for ensuring safety and quality of food and feed products, for preventing health risks and for avoiding economic losses because of product rejection or costly mycotoxin removal. A LED-based pocket-size fluorometer is presented that allows a rapid and low-cost screening of DON-contaminated durum wheat bran samples, without using chemicals or product handling. Forty-two samples with DON contamination in the 40-1650 µg/kg range were considered. A chemometric processing of spectroscopic data allowed distinguishing of samples based on their DON content using a cut-off level set at 400 µg/kg DON. Although much lower than the EU limit of 750 µg/kg for wheat bran, this cut-off limit was considered useful whether accepting the sample as safe or implying further inspection by means of more accurate but also more expensive standard analytical techniques. Chemometric data processing using Principal Component Analysis and Quadratic Discriminant Analysis demonstrated a classification rate of 79% in cross-validation. To the best of our knowledge, this is the first time that a pocket-size fluorometer was used for DON screening of wheat bran.

Optimizing extraction solvents for deoxynivalenol analysis in maize via infrared attenuated total reflection spectroscopy and chemometric methods

Farmers, cereal suppliers and processors demand rapid techniques for the assessment of mould-associated contamination. Deoxynivalenol (DON) is among the most important Fusarium toxins and related to human and animal diseases besides causing significant economic losses. Routine analytical techniques for the analysis of DON are either based on chromatographic or immunoanalytical techniques, which are time-consuming and frequently rely on hazardous consumables. The present study evaluates the feasibility of infrared attenuated total reflection spectroscopy (IR-ATR) for the analysis of maize extracts via different solvents optimized for the determination of DON contamination along the regulatory requirements by the European Union (EU) for unprocessed maize (1750 μg kg^-1). Reference analysis was done by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). The studied maize samples were either naturally infected or had been artificially inoculated in the field with Fusarium graminearum, Fusarium culmorum or Fusarium verticillioides. Principal component analysis demonstrated that water and methanol-water (70 : 30% v) were optimum solvents for differentiating DON contamination levels. Supervised partial least squares discriminant analysis resulted in excellent classification accuracies of 86.7% and 90.8% for water and methanol-water extracts, respectively. The IR spectra of samples with fungal infection and high DON contamination had distinct spectral features, which could be related to carbohydrates, proteins and lipid content within the investigated extracts.

The potential and applicability of infrared spectroscopic methods for the rapid screening and routine analysis of mycotoxins in food crops

Infrared (IR) spectroscopy is increasingly being used to analyze food crops for quality and safety purposes in a rapid, nondestructive, and eco-friendly manner. The lack of sensitivity and the overlapping absorption characteristics of major sample matrix components, however, often prevent the direct determination of food contaminants at trace levels. By measuring fungal-induced matrix changes with near IR and mid IR spectroscopy as well as hyperspectral imaging, the indirect determination of mycotoxins in food crops has been realized. Recent studies underline that such IR spectroscopic platforms have great potential for the rapid analysis of mycotoxins along the food and feed supply chain. However, there are no published reports on the validation of IR methods according to official regulations, and those publications that demonstrate their applicability in a routine analytical set-up are scarce. Therefore, the purpose of this review is to discuss the current state-of-the-art and the potential of IR spectroscopic methods for the rapid determination of mycotoxins in food crops. The study critically reflects on the applicability and limitations of IR spectroscopy in routine analysis and provides guidance to non-spectroscopists from the food and feed sector considering implementation of IR spectroscopy for rapid mycotoxin screening. Finally, an outlook on trends, possible fields of applications, and different ways of implementation in the food and feed safety area are discussed.

Electronic Nose for the Rapid Detection of Deoxynivalenol in Wheat Using Classification and Regression Trees

Mycotoxin represents a significant concern for the safety of food and feed products, and wheat represents one of the most susceptible crops. To manage this issue, fast, reliable, and low-cost test methods are needed for regulated mycotoxins. This study aimed to assess the potential use of the electronic nose for the early identification of wheat samples contaminated with deoxynivalenol (DON) above a fixed threshold. A total of 214 wheat samples were collected from commercial fields in northern Italy during the periods 2014−2015 and 2017−2018 and analyzed for DON contamination with a conventional method (GC-MS) and using a portable e-nose “AIR PEN 3” (Airsense Analytics GmbH, Schwerin, Germany), equipped with 10 metal oxide sensors for different categories of volatile substances. The Machine Learning approach “Classification and regression trees” (CART) was used to categorize samples according to four DON contamination thresholds (1750, 1250, 750, and 500 μg/kg). Overall, this process yielded an accuracy of >83% (correct prediction of DON levels in wheat samples). These findings suggest that the e-nose combined with CART can be an effective quick method to distinguish between compliant and DON-contaminated wheat lots. Further validation including more samples above the legal limits is desirable before concluding the validity of the method.

Infrared Spectroscopy–Quo Vadis?

Given the exquisite capability of direct, non-destructive label-free sensing of molecular transitions, IR spectroscopy has become a ubiquitous and versatile analytical tool. IR application scenarios range from industrial manufacturing processes, surveillance tasks and environmental monitoring to elaborate evaluation of (bio)medical samples. Given recent developments in associated fields, IR spectroscopic devices increasingly evolve into reliable and robust tools for quality control purposes, for rapid analysis within at-line, in-line or on-line processes, and even for bed-side monitoring of patient health indicators. With the opportunity to guide light at or within dedicated optical structures, remote sensing as well as high-throughput sensing scenarios are being addressed by appropriate IR methodologies. In the present focused article, selected perspectives on future directions for IR spectroscopic tools and their applications are discussed. These visions are accompanied by a short introduction to the historic development, current trends, and emerging technological opportunities guiding the future path IR spectroscopy may take. Highlighted state-of-the art implementations along with novel concepts enhancing the performance of IR sensors are presented together with cutting-edge developments in related fields that drive IR spectroscopy forward in its role as a versatile analytical technology with a bright past and an even brighter future.

Machine Learning for Predicting Mycotoxin Occurrence in Maize

Meteorological conditions are the main driving variables for mycotoxin-producing fungi and the resulting contamination in maize grain, but the cropping system used can mitigate this weather impact considerably. Several researchers have investigated cropping operations' role in mycotoxin contamination, but these findings were inconclusive, precluding their use in predictive modeling. In this study a machine learning (ML) approach was considered, which included weather-based mechanistic model predictions for AFLA-maize and FER-maize [predicting aflatoxin B1 (AFB1) and fumonisins (FBs), respectively], and cropping system factors as the input variables. The occurrence of AFB1 and FBs in maize fields was recorded, and their corresponding cropping system data collected, over the years 2005-2018 in northern Italy. Two deep neural network (DNN) models were trained to predict, at harvest, which maize fields were contaminated beyond the legal limit with AFB1 and FBs. Both models reached an accuracy >75% demonstrating the ML approach added value with respect to classical statistical approaches (i.e., simple or multiple linear regression models). The improved predictive performance compared with that obtained for AFLA-maize and FER-maize was clearly demonstrated. This coupled to the large data set used, comprising a 13-year time series, and the good results for the statistical scores applied, together confirmed the robustness of the models developed here.

Infrared attenuated total reflection spectroscopic surface analysis of bovine-tail intervertebral discs after UV-light-activated riboflavin-induced collagen cross-linking

The tensile strength of the intervertebral disc (IVD) is mainly maintained by collagen cross-links. Loss of collagen cross-linking combined with other age-related degenerative processes contributes to tissue weakening, biomechanical failure, disc herniation and pain. Exogenous collagen cross-linking has been identified as an effective therapeutic approach for restoring IVD tensile strength. The current state-of-the-art method to assess the extent of collagen cross-linking in tissues requires destructive procedures and high-performance liquid chromatography. In this study, we investigated the utility of infrared attenuated total reflection (IR-ATR) spectroscopy as a nondestructive analytical strategy to rapidly evaluate the extent of UV-light-activated riboflavin (B2)-induced collagen cross-linking in bovine IVD samples. Thirty-five fresh bovine-tail IVD samples were equally divided into five treatment groups: (a) untreated, (b) cell culture medium Dulbecco's Modified Eagle's Medium only, (c) B2 only, (d) UV-light only and (e) UV-light-B2. A total of 674 measurements have been acquired, and were analyzed via partial least squares discriminant analysis. This classification scheme unambiguously identified individual classes with a sensitivity >91% and specificity >92%. The obtained results demonstrate that IR-ATR spectroscopy reliably differentiates between different treatment categories, and promises an excellent tool for potential in vivo, nondestructive and real-time assessment of exogenous IVD cross-linking.