A matching model based on earth mover's distance for tracking Myxococcus xanthus

Area of Science:

• Microbiology
• Computational Biology
• Image Analysis

Background:

• Accurate tracking of bacterial cell motion is vital for fundamental biological research.
• Existing cell tracking methods struggle with complex behaviors like cell division and segmentation errors.

Purpose of the Study:

• To develop a robust and efficient algorithm for tracking bacterial cell motion, specifically Myxococcus xanthus.
• To overcome limitations of current detection association algorithms in handling segmentation inaccuracies and dynamic cell behaviors.

Main Methods:

• Proposed a polynomial time hierarchical approach for cell detection association.
• Introduced a novel Earth Mover's Distance (EMD) based matching model for improved cell segmentation association.
• Validated the method on challenging datasets involving Myxococcus xanthus.

Main Results:

• The new algorithm effectively tracks cell motion despite cell division, entry/exit from the image, and segmentation errors (false alarms, lost detections, merged/split detections).
• Demonstrated superior performance on error-prone segmented cells compared to state-of-the-art algorithms.
• Achieved higher track purity and more complete trajectories.

Conclusions:

• The developed EMD-based hierarchical approach offers a significant advancement in bacterial cell tracking.
• This method provides a more reliable solution for analyzing complex bacterial populations in biological studies.
• Enhances the accuracy and completeness of bacterial motion trajectory analysis.