Differentiation of toxigenic fungi using hyperspectral imagery

A Vis/NIR spectra-based approach for identifying bananas infected with Colletotrichum musae

Introduction

Anthracnose of banana caused by Colletotrichum species is one of the most serious post-harvest diseases, which can cause significant yield losses. Clarifying the infection mechanism of the fungi using non-destructive methods is crucial for timely discriminating infected bananas and taking preventive and control measures.

Methods

This study presented an approach for tracking growth and identifying different infection stages of the C. musae in bananas using Vis/NIR spectroscopy. A total of 330 banana reflectance spectra were collected over ten consecutive days after inoculation, with a sampling rate of 24 h. The four-class and five-class discriminant patterns were designed to examine the capability of NIR spectra in discriminating bananas infected at different levels (control, acceptable, moldy, and highly moldy), and different time at early stage (control and days 1-4). Three traditional feature extraction methods, i.e. PC loading coefficient (PCA), competitive adaptive reweighted sampling (CARS) and successive projections algorithm (SPA), combining with two machine learning methods, i.e. partial least squares discriminant analysis (PLSDA) and support vector machine (SVM), were employed to build discriminant models. One-dimensional convolutional neural network (1D-CNN) without manually extracted feature parameters was also introduced for comparison.

Results

The PCA-SVM and·SPA-SVM models had good performance with identification accuracies of 93.98% and 91.57%, 94.47% and 89.47% in validation sets for the four- and five-class patterns, respectively. While the 1D-CNN models performed the best, achieving an accuracy of 95.18% and 97.37% for identifying infected bananas at different levels and time, respectively.

Discussion

These results indicate the feasibility of identifying banana fruit infected with C. musae using Vis/NIR spectra, and the resolution can be accurate to one day.

Fungi Classification in Various Growth Stages Using Shortwave Infrared (SWIR) Spectroscopy and Machine Learning

Dark septate endophytes (DSEs) fungi are beneficial to host plants with regard to abiotic stress. Here, we examined the capability of SWIR spectroscopy to classify fungus types and detected the growth stages of DSEs fungi in a timely, non-destructive and time-saving manner. The SWIR spectral data of five DSEs fungi in six growth stages were collected, and three pre-processing methods and sensitivity analysis (SA) variable selection methods were performed using a machine learning model. The results showed that the De-trending + first Derivative (DET_FST) processing spectra combined with the support vector machine (SVM) model yielded the best classification accuracy for fungi classification at different growth stages and growth stage detection on different fungus types. The mean accuracy of generic model for fungi classification and growth stage detection are 0.92 and 0.99 on the calibration set, respectively. Seven important bands, 1164, 1456, 2081, 2272, 2278, 2448 and 2481 nm, were found to be related to the SVM fungi classification. This study provides a rapid and efficient method for the classification of fungi in different growth stages and the detection of fungi growth stage of various types of fungi and could serve as a tool for fungi study.

Application of near-infrared hyperspectral (NIR) images combined with multivariate image analysis in the differentiation of two mycotoxicogenic Fusarium species associated with maize

Maize (Zea mays L.) is one of the most versatile crops worldwide with high socioeconomic relevance. However, mycotoxins produced by pathogenic fungi are of constant concern in maize production, as they pose serious risks to human and animal health. Thus, the search for rapid detection and/or identification methods for mycotoxins and mycotoxin-producing fungi for application in food safety remain important. In this work, we implemented use of near infrared hyperspectral images (HSI-NIR) combined with pattern recognition analysis, partial-least-squares discriminant analysis (PLS-DA) of images, to develop a rapid method for identification of Fusarium verticillioides and F. graminearum. Validation of the HSI-NIR method and subsequent analysis was realized using 15 Fusarium spp. isolates. The method was efficient as a rapid, non-invasive, and non-destructive assessment was achieved with 100% accuracy, sensitivity, and specificity for both fungi.

Development of Methods for Determination of Aflatoxins

Aflatoxins can cause damage to the health of humans and animals. Several institutions around the world have established regulations to limit the levels of aflatoxins in food, and numerous analytical methods have been extensively developed for aflatoxin determination. This review covers the currently used analytical methods for the determination of aflatoxins in different food matrices, which includes sampling and sample preparation, sample pretreatment methods including extraction methods and purification methods of aflatoxin extracts, separation and determination methods. Validation for analysis of aflatoxins and safety considerations and precautions when doing the experiments are also discussed.

Growth Simulation and Discrimination of Botrytis cinerea, Rhizopus stolonifer and Colletotrichum acutatum Using Hyperspectral Reflectance Imaging

This research aimed to develop a rapid and nondestructive method to model the growth and discrimination of spoilage fungi, like Botrytis cinerea, Rhizopus stolonifer and Colletotrichum acutatum, based on hyperspectral imaging system (HIS). A hyperspectral imaging system was used to measure the spectral response of fungi inoculated on potato dextrose agar plates and stored at 28°C and 85% RH. The fungi were analyzed every 12 h over two days during growth, and optimal simulation models were built based on HIS parameters. The results showed that the coefficients of determination (R²) of simulation models for testing datasets were 0.7223 to 0.9914, and the sum square error (SSE) and root mean square error (RMSE) were in a range of 2.03–53.40×10⁻⁴ and 0.011–0.756, respectively. The correlation coefficients between the HIS parameters and colony forming units of fungi were high from 0.887 to 0.957. In addition, fungi species was discriminated by partial least squares discrimination analysis (PLSDA), with the classification accuracy of 97.5% for the test dataset at 36 h. The application of this method in real food has been addressed through the analysis of Botrytis cinerea, Rhizopus stolonifer and Colletotrichum acutatum inoculated in peaches, demonstrating that the HIS technique was effective for simulation of fungal infection in real food. This paper supplied a new technique and useful information for further study into modeling the growth of fungi and detecting fruit spoilage caused by fungi based on HIS.

Developments in detection and determination of aflatoxins

Since the discovery of aflatoxins in the 1960s, much research has focused on detecting the toxins in contaminated food and feedstuffs in the interest of public safety. Most traditional detection methods involved lengthy culturing and/or separation techniques or analytical instrumentation and complex, multistep procedures that required destruction of samples for accurate toxin determination. With more regulations for acceptable levels of aflatoxins in place, modern analytical methods have become quite sophisticated, capable of achieving results with very high precision and accuracy, suitable for regulatory laboratories and for post-harvest sample testing in developed countries. Unfortunately, many countries around the world that are affected by the aflatoxin problem do not have ready access to high performance liquid chromatography and mass spectrometry instrumentation and require alternate, readily available and simple detection methods that may be used by small holdings farmers in developing countries. This paper presents an overview of the existing detection and/or determination methods for aflatoxins. The traditional, quantitative, chemically-based analytical strategies for detecting aflatoxins in maize and their evolution to the modern instrumentation routinely used in developed countries are reviewed. Additionally, novel, more streamlined, user-friendly and in some instances, non-destructive, methods that may be useful for semi-quantitative or qualitative, quick-screening of contaminated maize samples appropriate also for use in developing countries, are discussed.

Near-infrared (NIR) hyperspectral imaging and multivariate image analysis to study growth characteristics and differences between species and strains of members of the genus Fusarium

Near-infrared (NIR) hyperspectral imaging was used to study three strains of each of three Fusarium spp. (Fusarium subglutinans, Fusarium proliferatum and Fusarium verticillioides) inoculated on potato dextrose agar in Petri dishes after either 72 or 96 h of incubation. Multivariate image analysis was used for cleaning the images and for making principal component analysis (PCA) score plots and score images and local partial least squares discriminant analysis (PLS-DA) models. The score images, including all strains, showed how different the strains were from each other. Using classification gradients, it was possible to show the change in mycelium growth over time. Loading line plots for principal component (PC) 1 and PC2 explained variation between the different Fusarium spp. as scattering and chemical differences (protein production), respectively. PLS-DA prediction results (including only the most important strain of each species) showed that it was possible to discriminate between species with F. verticillioides the least correctly predicted (between 16 and 47 % pixels correctly predicted). For F. subglutinans, 78-100 % pixels were correctly predicted depending on the training and test sets used. Similarly, the percentage correctly predicted values of F. proliferatum were 60-80 %. Visualisation of the mycelium radial growth in the PCA score images was made possible due to the use of NIR hyperspectral imaging. This is not possible with bulk spectroscopy in the visible or NIR regions.