Image-Based Quantification of Arabidopsis thaliana Stomatal Aperture from Leaf Images.

The article focuses on the development of an automated image analysis pipeline combined with a portable imaging device to quantify stomatal aperture in Arabidopsis thaliana leaves from bright-field microscopy images, including those with noisy mesophyll backgrounds. The pipeline integrates a You Only Look Once X (YOLOX)–based stomatal detection module and a U-Net–based pore segmentation module, achieving high precision in detecting stomata and segmenting pores, with a quantification error of approximately 0.2 µm compared to manual measurements. The portable device enables nondestructive, on-site imaging of both detached and intact leaves, and when paired with a fine-tuned model, it accurately reproduces stomatal responses to bacterial inoculation while significantly reducing analysis time. This system offers a practical tool to facilitate high-throughput and precise physiological studies of stomatal dynamics in Arabidopsis under various environmental and biotic conditions.