Optical Coherence Tomography for Monitoring Plants’ Responses to Biotic Stressors

This project explores Optical Coherence Tomography (OCT) as a potential new imaging technique to detect early signs of infection in plants. OCT can be applied to deliver cross-sectional and three-dimensional images of plant’s microstructures non-invasively, in vivo, and in real-time. This puts forward a very exciting potential methodology for boosting our understanding of plant’s response to stressors, plant health, and disease management. OCT traditionally finds wide application in medical imaging, particularly in ophthalmology, to image the retina with high resolution. This research project aims to adapt this technology to the realm of botany. More specifically, the objective is to monitor the response of wheat plants when infected by septoria. In this aim, OCT provides detailed images of inner structures without affecting the plant, which enables continuous monitoring without altering the growth and development of the plant. Specifically, the study monitors the response of wheat plants infected by Septoria. This project thus corresponds to a proof-of-concept which demonstrate the detection of early signs of infection before any external signs are visible.
The principal strength of OCT lies in its ability to provide detailed internal images without damaging the plant, allowing continuous monitoring of growth and disease progression. This non-invasive capacity is crucial for pathogen detection, which in turn maximizes crop survival and minimizes chemical treatment use. In my study, OCT was benchmarked against advanced techniques such as confocal microscopy, epifluorescence microscopy, and scanning electron microscopy (SEM), alongside manual analysis using FIJI and MATLAB, confirming its reliability.
However, OCT also presents certain limitations. Compared with established imaging methods, its penetration depth in dense plant tissues is limited, which can result in blurry images. In addition, interpretation of OCT images requires specialized expertise and dedicated software to extract meaningful data. In this study, collaboration with a software company enabled the development of customized machine learning based segmentation software. This tool allows rapid, automated OCT image analysis, producing immediate quantitative results and significantly enhancing the efficiency of plant research.
The OCT analyser software, based on machine learning (ML), applies masks to the inner structure of leaves to enable gap thickness measurements and extract meaningful data. However, the current version of the software was trained on only one variety (AxC169). To improve its robustness, it will need to be trained on additional varieties. The gap thickness measurements, whether performed manually or through automated approaches, are provided in this study to ensure availability for other researchers.
Overall, the integration of OCT with ML is highly promising and has the potential to open new opportunities for agricultural applications