Rapid and Non-Destructive Detection of Aflatoxin in Post-harvest Peanuts Using Advanced Imaging Approaches for Enhanced Food Safety

Commodities

Practices

Aflatoxins (AFs) are toxic fungal metabolites produced by Aspergillus flavus and Aspergillus parasiticus, posing a major global concern. AFs pose serious health risks, including liver cancer, weakened immunity, and childhood stunting, and they also cause significant economic losses globally. Detecting aflatoxins in post-harvest peanuts is essential for ensuring food safety, producing safe crops, and meeting regulatory standards for allowable aflatoxin levels. Currently, chemical testing methods such as mass spectrometry, gas chromatography, thin-layer chromatography, and high-performance liquid chromatography are used to monitor for aflatoxins along the value chain. However, these methods are destructive, time-consuming, expensive, and unsuitable for large-scale screening and removal of contaminated seeds in production environments. To address these challenges, the long-term objective of this project is to develop an autonomous screening system capable of distinguishing and separating aflatoxin-contaminated peanuts from non-contaminated ones. As an initial step towards the long-term goal, the current project focuses on developing a rapid, non-destructive and environmentally friendly detection system for aflatoxin B1 on peanut surfaces using advanced imaging technologies, such as hyperspectral and ultraviolet fluorescence imaging. Findings from this study will be integrated into an autonomous screening line to automate the detection and removal of aflatoxin-contaminated peanuts. This advancement will enable real-time, high-throughput screening in post-harvest processing, significantly improving food safety and reducing economic losses. Ultimately, the project aims to support a safer, more efficient, and sustainable peanut supply chain.

The primary objective of this project is to develop rapid, non-destructive, and cost-effective methods for detecting aflatoxin contamination in post-harvest peanut kernels using advanced imaging technologies. The study focuses on two key modalities: hyperspectral imaging and ultraviolet fluorescence imaging. Hyperspectral imaging, also known as imaging spectroscopy, is an emerging tool for detecting aflatoxins. By capturing hundreds of spectral bands, it can detect subtle differences in photon absorption caused by variations in molecular structure. These spectral features enable qualitative assessment of aflatoxin presence on the surface of peanut kernels (Gao et al., 2021b). Ultraviolet (UV) fluorescence imaging is another powerful technique that can be useful for detecting aflatoxins based on their natural fluorescence properties. When exposed to UV light, aflatoxins emit characteristic fluorescence signals that can be captured and analyzed to identify contaminated areas on peanut surfaces. This method enables rapid, non-contact screening and is particularly effective for highlighting surface-level contamination with high contrast and sensitivity. This project will focus on two primary objectives: comparing the performance of two imaging modalities and identifying the most effective method for detecting aflatoxin B1 in post-harvest peanuts.

Objective 1: Application of hyperspectral imaging and machine learning for detecting aflatoxin B1 in post-harvest peanuts above regulatory limits.

This objective will utilize near-infrared (NIR) and visible-near-infrared (VNIR) hyperspectral imaging to detect aflatoxin contamination in peanut kernels. The process involves collecting hyperspectral data, extracting relevant spectral features, and applying machine learning and deep learning models to classify contamination levels. Key wavelength bands will be identified using dimensionality reduction techniques such as Principal Component Analysis (PCA) and Competitive Adaptive Reweighted Sampling (CARS). These selected features will then be used to train models to accurately detect peanuts contaminated with aflatoxin B1.

Objective 2. Utilize ultraviolet fluorescence imaging combined with feature analysis and machine learning to detect aflatoxin B1 in post-harvest peanuts.

This objective will utilize ultraviolet (UV) fluorescence imaging to detect aflatoxin B1 contamination in post-harvest peanut kernels. The approach involves capturing fluorescence emissions using a portable imaging system equipped with UV illumination and an RGB camera. Bright green-yellow fluorescence, a known indicator of aflatoxin presence, will be quantified using image-based feature extraction techniques. The acquired data will be processed and analyzed using image processing and machine learning approaches for classifying contaminated peanut kernels.