Mycotoxin prediction with hyperspectral imaging
Mycotoxin contamination in small grain cereals can lead to safety incidents, animal and human health problems and economic losses. Mycotoxin production is largely influenced by weather conditions during critical crop growing stages (i.e. flowering and harvesting) and agricultural management practices (i.e. soil tillage, cultivar, previous crop, fungicide application, etc). Due to the temperature increase and changing precipitation pattern changes caused by climate change,
mycotoxin contamination in food and feed crops has become one of the top threats worldwide for human and animal health. A reliable precise mycotoxin early warning system is therefore needed to ensure food and feed safety without polluting the environment and to achieve a sustainable agriculture.
The proposed project aims to predict on-site mycotoxin contamination in cereal grains in the Netherlands at early crop growing stage using new technologies like hyperspectral imaging (HSI) and machine learning. Such a site-specific crop management system will help farmers to do the right thing in the right place, in the right way, at the right time. For instance, based on predictions of the presence of mycotoxin producing fungi, farmers can use fungicides in the right dose
exactly at places in the field where it is needed. This multi-disciplinary project will integrate a variety of data, farm expert knowledge, technologies and algorithms with a dynamic consortium of farmers, farm cooperation, collectors, software developers, and researchers on phenotyping, precision farming and food safety. Ultimately, the proposed work will further improve and validate the existing mycotoxin prediction model ‘DON-Control Tarwe’ with more detailed field
monitoring data, optimize the application of hyper/multi-spectral imaging technique in the field condition, and explore the possibility of Fusarium spp. early detection, in order to increase the fungicide use efficiency and limit mycotoxin contamination in the Netherlands.
The sector aims to produce and consume healthy, safe, high quality and sustainable food with farmers earning fair prices in 2030. Specifically, the sector has prioritized to develop methods and systems for the early identification of chemical and microbiological food safety hazards in the food chain, in particular in the area of rapid detection methods for food safety control. To limit the impacts of mycotoxins on economic losses and health problems, on-site monitoring and
detection is the key for farmers and advisors to be informed on time.
Mycotoxin contamination needs be prevented and controlled in the field by various management practices, such as fungicides spraying against Fusarium Head Blight. However, these fungicides should be applied in time and with care. A dedicated, timely and spatial specific fungicide spraying plan and management advice are needed to limit mycotoxin contamination in small grain cereals and to contribute to a green and sustainable agriculture in the Netherlands.
The results of this project will deliver for the first time an early warning system for mycotoxin prediction using innovative Fusarium detection techniques, in particular HSI, as inputs, and making use of machine learning in system development. Such a decision support system will identify high-risk areas with fungal infection in the field (hot spots) and provide information to farmers on when and where to spray fungicides to limit the contamination.
Detailed timeline and deliverables are listed in table below.
D1.1 Yearly multi-mycotoxin analysis results sent to the growers (Year 1)
D1.2 Tested web application DON-Control Tarwe for DON prediction (Year 2)
D1.3 Functioning DON-Control Mais web application (Year 3)
D2.1 Detailed four years experimental design (Year 1)
D2.2 Algorithm to detect the presence of Fusarium spp. at early growing stage (Year 3)
D2.3 Updated mycotoxin prediction model (Year 4)
D3.1 Signed consortium agreement submitted to TKI (Year 1)
D3.2 Yearly dissemination activities (stakeholder workshop to demonstrate Don-Control Tarwe, conference, peer-review publications, news articles, etc)
Het e-mailadres wordt niet gepubliceerd. Vereiste velden zijn gemarkeerd met *