Cell Tracking According to Biological Needs-Strong Mitosis-Aware Multi- Hypothesis Tracker With Aleatoric...

Area of Science:

• Cell Biology
• Bioimage Analysis
• Computational Biology

Background:

• Accurate cell tracking and segmentation are crucial for analyzing microscopy time-lapse data.
• Current methods struggle with long-term consistency and correct lineage tree reconstruction due to reliance on local accuracy metrics.

Purpose of the Study:

• To develop a novel cell tracking assignment strategy that enhances long-term consistency and lineage reconstruction.
• To address limitations in current tracking approaches for large-scale microscopy data.

Main Methods:

• Introduced an uncertainty estimation technique for motion estimation, relaxing single-point representations into probabilistic spatial densities.
• Developed a novel mitosis-aware assignment problem formulation leveraging spatial densities to model cell divisions and resolve conflicts.
• Integrated biological knowledge with learned representations derived from spatial densities.

Main Results:

• The proposed framework significantly outperforms the state-of-the-art on biologically inspired metrics across nine datasets.
• Achieved improvements by a factor of approximately six in cell tracking accuracy and lineage reconstruction.
• Provided new insights into the behavior and impact of motion estimation uncertainty.

Conclusions:

• The novel assignment strategy enhances cell tracking accuracy and long-term consistency.
• The mitosis-aware formulation effectively models cell divisions and improves lineage tree reconstruction.
• This approach offers a significant advancement for analyzing complex biological microscopy data.