Abstract
Arabidopsis is a widely used model plant to study physiology and development. Live imaging is an important technique to visualize and quantify processes in plant growth and cell division, where accurate cell tracking is essential. The commonly used software TrackMate adopts a tracking-by-detection approach, applying Laplacian of Gaussian (LoG) for blob detection and a Linear Assignment Problem (LAP) tracker for tracking. However, its performance declines when cells are densely arranged. To overcome this limitation, we propose an accurate tracking method based on a Genetic Algorithm (GA) that incorporates knowledge of Arabidopsis root cellular patterns and spatial relationships among volumes. Our method follows a coarse-to-fine strategy: first performing relatively simple line-level tracking of nuclei, then refining associations based on the linear arrangement of cell files and their spatial relationships. We evaluated the method on long-term live imaging datasets of Arabidopsis root tips, and with minor manual correction, it achieved accurate nuclear tracking. To the best of our knowledge, this represents the first successful attempt to address a long-standing problem in time-lapse microscopy of the root meristem by providing an accurate tracking method for Arabidopsis root nuclei.
Introduction
Understanding how plants grow requires capturing the "life story" of individual cells: when they divide, how fast they elongate, and how they move. While 3D time-lapse microscopy provides the raw data, automated tracking remains a bottleneck in the Arabidopsis root meristem. Standard tools like TrackMate often rely on the "Linear Assignment Problem" (LAP), which links objects between frames based primarily on Euclidean distance. This approach frequently fails in the root tip due to high cellular density, stochastic rotational movement, and the visual similarity of nuclei.
Methods: Bio-Informed Tracking Pipeline
The proposed framework moves from global structures to individual cells by leveraging the biological fact of the root's natural eight-file symmetry.
The pipeline consists of several key stages:
- Deep Learning Detection: A U-Net architecture performs three-class segmentation to identify background, non-dividing nuclei, and dividing (mitotic) nuclei.
- Genetic Algorithm (GA) Projection: To handle root rotation, a GA is employed to find the optimal projection plane. The fitness function minimizes the standard deviation of distances between nuclei and their centroids, searching for the angle where the eight cell files are most clearly separated.
- K-Means Clustering: Once the optimal plane is found, nuclei are clustered into eight distinct groups representing the cell files.
- Line-Level Tracking: The algorithm first tracks the eight lines using polar coordinates. Linking lines based on minimal angular displacement is significantly more stable than Cartesian distance during root rotation.
- Nucleus-Level Refinement: Within each tracked line, individual nuclei are linked based on their vertical position and division status.
Results and Performance
The study utilized three high-resolution 3D time-lapse datasets of Arabidopsis roots with RFP-labeled nuclei, spanning up to 144 time points at 15-30 minute intervals.
The GA-based method demonstrated superior accuracy compared to traditional approaches. While the TrackMate LAP tracker achieved ~51.4% recall, the proposed method achieved 84.6% precision and 71.3% recall without human intervention. With minor manual refinement of the initial K-means clustering, the method reached 100% tracking accuracy across 144 frames. Furthermore, the GA-optimized projection achieved a clustering accuracy of 99.53%, vastly outperforming standard techniques like Principal Component Analysis (PCA).
Conclusion
By integrating biological priors into a Genetic Algorithm-based framework, this research provides a robust solution for the long-standing problem of tracking nuclei in the dense Arabidopsis root meristem. While the current implementation is tuned to the eight-file symmetry of the cortex, it establishes a "coarse-to-fine" paradigm that could be adapted to other tissues or species by redefining structural priors. Future work will focus on reducing the need for manual refinement and improving computational efficiency.
Summary updated