Machine learning models for root image analysis in wheat nitrogen use efficiency research
Research collaboration opportunity using machine learning to analyze root images from wheat field trials to identify nitrogen fertilizer regimes and assess nitrogen use efficiency across genotypes with contrasting senescence traits.
Overview
This research initiative, led by the University of Lincoln, applies machine learning approaches to root image analysis in wheat to advance understanding of nitrogen use efficiency (NUE). The study examines 25 wheat genotypes with contrasting senescence characteristics grown under two nitrogen fertilizer rates across three replicates. The core hypothesis is that root length will be longer under low nitrogen conditions in lines exhibiting stay-green properties, and that these differences can be detected using machine learning classifiers trained on root imagery. The research aims to address the lack of high-throughput field phenotyping methods for root traits in breeding programs.
Technical specifications
Research approach:
- Field trial drilled in October 2021 at Riseholme with 25 genotypes contrasting for senescence at two nitrogen fertilizer rates over three replicates
- In-situ root scanning using a mini-rhizotron, an in-house sensor available at Lincoln, capturing high-resolution digital images of living root systems
- Clear acrylic tubes inserted in all 150 plots to enable non-destructive root observation throughout the growing season
- Weekly measurements taken during the growing season to track root development dynamics
- Machine learning classifier construction from root images to accurately identify which fertilizer regime the plant was grown under
- Investigation of root architecture as a determinant of nitrogen uptake efficiency
Technology readiness level
The experiment commenced in October 2021 and the hypothesis has not yet been validated. The research team is seeking funding for the field acrylic tubes and for the postdoctoral researcher who will build the machine learning models. Future validation will focus on constructing the machine learning classifier from collected root images to test whether root systems produced under different nitrogen regimes can be reliably distinguished. This work represents an early-stage research collaboration opportunity with potential applications in crop breeding and nitrogen management.
About University of Lincoln
The University of Lincoln is a comprehensive public university with a modern, city-centre campus and an applied research culture. Industry engagement is built into its model through co-located facilities at a dedicated science and innovation park and long-term partnerships with global manufacturers, utilities, and regional SMEs. Specialized food manufacturing and packaging testbeds at its south Lincolnshire campus provide pilot-scale lines for process development, while links with regional NHS trusts and a medical school enable clinical research and evaluation. Research draws competitive support from UKRI councils, Innovate UK, and the National Institute for Health and Care Research, alongside partnerships with industry. A dedicated technology transfer and enterprise office supports IP management, licensing, and spinout formation.
delivered to your inbox.