BACKGROUND

Fungal diseases are among the most significant threats to global crop production, often leading to substantial yield losses. Early detection of crop infection by fungus is the very first step to deploying a timely and effective treatment. Early and reliable detection is thus key to improving yields, sustainability, and achieving food security. Conventional diagnostic methods are however often destructive, slow, or requiring visible symptoms which appear late in the infection process. To overcome these challenges, we propose using optical coherence tomography (OCT) as an innovative imaging tool to provide cross-sectional and three-dimensional images of the plant internal microstructure non-invasively, in vivo, and in real-time.

RESULTS

We demonstrate the use of low-cost OCT to monitoring wheat (cultivar AxC 169) when infected by Septoria tritici. We show that OCT analysis can effectively detect signs of infection before any external symptoms appear. Although OCT cannot directly visualize fungal hyphae, OCT reveals apparent morphological changes of the mesophyll where the fungal filaments are expected to develop. This study thus focuses on monitoring and correlating changes within the mesophyll structural organisation with the state of infection. It results in distinct statistical difference between intact and infected wheat plants two days only after infection. We then demonstrate the use of machine learning (ML) for high throughput segmentation of OCT scans, providing a foundation for future automated fungus-detection analysis.

CONCLUSIONS

This work highlights the potential of OCT, combined with ML tools, to enable rapid, non-invasive, and early diagnosis of crop fungal infections, opening new avenues for precision agriculture and sustainable disease management.