Multi-Target Tracking With Time-Varying Clutter Rate and Detection Profile: Application to Time-Lapse Cell Microscopy...

Area of Science:

• Cellular and Molecular Biology
• Biophysics
• Image Analysis

Background:

• Accurate tracking of cellular and sub-cellular particles in microscopy is crucial for understanding cell dynamics.
• Existing multi-target tracking methods struggle with high noise, dense populations, complex motion, and changing image conditions.

Purpose of the Study:

• To develop a robust framework for multi-target particle tracking in challenging time-lapse cell microscopy sequences.
• To address limitations of current methods in handling noise, missed detections, and spurious measurements.

Main Methods:

• A novel framework based on the random finite set Bayesian filtering approach is proposed.
• A bootstrap filter, comprising an adaptive estimator and a state tracker, is utilized.
• The estimator dynamically adjusts key parameters like clutter rate and detection probability.

Main Results:

• The proposed framework demonstrates superior performance compared to state-of-the-art particle trackers.
• Effective tracking was achieved on both synthetic and real microscopy data with challenging image characteristics.
• The adaptive estimator successfully managed time-varying detection and clutter rates.

Conclusions:

• The developed Bayesian filtering framework offers a reliable solution for quantitative analysis of particle dynamics in microscopy.
• The adaptive estimation component enhances tracking robustness under adverse imaging conditions.
• This approach advances the capability for precise analysis of cellular processes from time-lapse imaging.