Tracking Arabidopsis growth from time-lapse images
Collaboration
F. Lambert, Department of Biology, Faculty of Science, University of Copenhagen
Research Background
The project aims to investigate if specific RNA methylations affect plant growth. Arabidopsis plants are seeded on a petri dish, and imaged from an aerial view using a standard digital camera every 24 hours, with a ruler in the image for scale. To see if the mutations cause plants to grow at different rates, we need to track each plant and measure their area over time.
Method
We developed a set of four ImageJ macros that form a multi-component pipeline for processing the images. The four macros are designed to automate the respective steps:
Image Alignment: The natural RGB images are converted into a greyscale image and processed with ImageJ’s Find Edges filter to detect its edges. We use the edge image to perform image registration with the Correct 3D Drift plugin, which is then used to align the original natural images.
Plant segmentation: The natural images are segmented into plant vs. background using the Labkit plugin. The training of the model is first done by the user on a subset of the data, and then applied in batch mode on every image.
Plant tracking: Plants from the previous segmentation step are identified via connected component labelling. Each identified plant is assigned to an object track, by comparing their overlap with a plant identified in the previous time point. Over time, as plants grow they may end up touching; touching plant labels are split by dilating the label of the previous time-point until they touch, and drawing the boundary there.
Measurement (with automatic scale detection): The final macro extracts the area of each segmented plant track over time in pixel units. The macro also automatically detects the real-world scale from a user-selected ROI of the ruler in the image, to convert the area measurements from pixel2 to mm2.
Impact
The data generated from this project were used to successfully apply for research funding (Novo Nordisk Foundation - Plant Science, Agriculture and Food Biotechnology 2025: An RNA methylation strategy for engineering plant tissue growth and increasing crop yields).
Code Repository
Will follow soon.