Using machine learning to predict plant environmental exposure
Lead Research Organisation:
Imperial College London
Department Name: Life Sciences
Abstract
Our ability to study and understand plant environmental response if predicated on our ability to measure and quantify plant physiological changes in response to varying conditions. The overall aim of this PhD is to develop novel methods to measure and quantify spectral reflectance changes in plant leaves to enable improved detection of subtle non-visible changes in plant reflectance under different plant stress conditions in lab and field trials.
During this PhD, I will be investigating the response of model plants and crop species to different stress conditions within lab and field experiments. This will involve the use of both hyperspectral and multispectral sensing to sample and image plant reflectance changes in response to nutrient and pathogen stress in monocot and dicot species. Machine learning classifiers will be built that enable distinction of plants exposed to different environmental conditions.
Controlled Environment Trials: Arabidopsis, tomato and wheat will be exposed to various biotic and abiotic stresses to investigate the impact of these stresses on plant leaf reflectance as captures by hyperspectral instrumentation and multispectral cameras. The spectra will then be analysed using multivariate analysis techniques to distinguish the regions in the spectra that change upon exposure to the different stresses.
Field Trials: Annually, Agrii (the industrial partner of this project) are planting a field trials using wheat. During these trials, controlled fungicide application will be used to investigate the progression of Septoria infection on wheat. The aim here will be to develop hyperspectral and multispectral systems that can reliably detect and monitor early asymptomatic disease progression in order to help better monitor and understand this important agricultural pathogen. Additionally, drones equipped with multispectral cameras will also be used to image trials to develop image classifiers capable of classifying and predicting Septoria infection.
Metabolomic and/or RNA-Seq profiling will be performed on selected samples to help understand how secondary metabolism and gene expression changes might be influencing plant leaf reflectance under different stress conditions.
The expected outcomes are: 1) A better understanding of how environmental exposure altered plant leaf reflectance and how this can be captured to better detect and quantify different stresses in the lab and in the field. 2) A multispectral imaging system to help with forecasting Septoria infection in the field. 3) Better understanding of the molecular networks operating in fields in response to complex stress, identified regulators as potential targets for future breeding or crop protection efforts
During this PhD, I will be investigating the response of model plants and crop species to different stress conditions within lab and field experiments. This will involve the use of both hyperspectral and multispectral sensing to sample and image plant reflectance changes in response to nutrient and pathogen stress in monocot and dicot species. Machine learning classifiers will be built that enable distinction of plants exposed to different environmental conditions.
Controlled Environment Trials: Arabidopsis, tomato and wheat will be exposed to various biotic and abiotic stresses to investigate the impact of these stresses on plant leaf reflectance as captures by hyperspectral instrumentation and multispectral cameras. The spectra will then be analysed using multivariate analysis techniques to distinguish the regions in the spectra that change upon exposure to the different stresses.
Field Trials: Annually, Agrii (the industrial partner of this project) are planting a field trials using wheat. During these trials, controlled fungicide application will be used to investigate the progression of Septoria infection on wheat. The aim here will be to develop hyperspectral and multispectral systems that can reliably detect and monitor early asymptomatic disease progression in order to help better monitor and understand this important agricultural pathogen. Additionally, drones equipped with multispectral cameras will also be used to image trials to develop image classifiers capable of classifying and predicting Septoria infection.
Metabolomic and/or RNA-Seq profiling will be performed on selected samples to help understand how secondary metabolism and gene expression changes might be influencing plant leaf reflectance under different stress conditions.
The expected outcomes are: 1) A better understanding of how environmental exposure altered plant leaf reflectance and how this can be captured to better detect and quantify different stresses in the lab and in the field. 2) A multispectral imaging system to help with forecasting Septoria infection in the field. 3) Better understanding of the molecular networks operating in fields in response to complex stress, identified regulators as potential targets for future breeding or crop protection efforts
People |
ORCID iD |
| Oliver Windram (Primary Supervisor) | |
| Pietro Spanu (Primary Supervisor) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| BB/M011178/1 | 01/10/2015 | 25/02/2025 | |||
| 2368380 | Studentship | BB/M011178/1 | 01/10/2016 | 30/09/2020 |