Tracking leukocytes in intravital time lapse images using 3D cell association learning network

Area of Science:

• Immunology
• Bioimaging
• Computational Biology

Background:

• Leukocytes are vital for innate immunity, but tracking their movement in vivo, like in zebrafish embryos, is difficult.
• Manual analysis of leukocyte migration data is time-consuming and laborious.
• Existing cell tracking methods struggle with accurately associating amorphous cells across image frames.

Purpose of the Study:

• To develop an automated method for tracking leukocyte migration in 3D in vivo imaging datasets.
• To address the challenge of cell association for amorphous and highly mobile cells.
• To improve the efficiency and accuracy of cellular immunology research.

Main Methods:

• A novel deep-learning-based object linkage method was proposed.
• A 3D cell association learning network (3D-CALN) was trained using manually labeled 3D image pairs of zebrafish neutrophils.
• The network was trained on consecutive frames to learn cell association.

Main Results:

• Deep learning proved effective for cell linkage and tracking amorphous, mobile leukocytes.
• The developed 3D-CALN demonstrated strong performance in cell association tasks.
• Comparative analysis showed the proposed method outperforms existing algorithms in cell tracking accuracy.

Conclusions:

• The novel deep learning approach significantly enhances the ability to track leukocyte migration in complex biological systems.
• This method offers a powerful tool for advancing research in cellular immunology and in vivo imaging.
• Automated cell tracking using deep learning is a promising solution for laborious manual analysis in immunology.