In the realm of developmental biology and advanced microscopy, capturing the dynamic journey of cells as they organize into complex tissues and organs has long presented a significant technical challenge. Although modern microscopes can now record striking, high-resolution videos of entire embryos maturing in real time, extracting meaningful data from these visualizations—specifically, reconstructing the trajectories of individual cells—is an intricate task fraught with difficulties. Cells move constantly, divide unpredictably, and sometimes even disappear from view, complicating efforts to monitor their paths accurately. Researchers rely on the nuclei within cells as key landmarks to delineate each cell’s boundaries, a procedure known as segmentation, followed by tracking these segmented entities frame by frame through the developmental sequences. Accurate cell tracking is indispensable, not only for understanding embryogenesis but also for revealing how diseases evolve and respond to treatments at a cellular level.
Recently, a team of scientists at the Chan Zuckerberg Biohub in San Francisco introduced Ultrack, a groundbreaking cell-tracking platform that dramatically enhances the precision and scalability of cell trajectory analysis. Published in the high-impact journal Nature Methods, Ultrack stands out for its ability to scale from tracking minimal numbers of cells under controlled laboratory conditions to analyzing entire embryos captured in complex three-dimensional (3D) microscopy videos. This platform demonstrated superior performance in the prestigious Cell Tracking Challenge, an international benchmark initiative designed to evaluate the efficacy of cell-tracking algorithms. By offering remarkable speed and adaptability, Ultrack offers scientists a powerful tool that pushes the frontiers of what is possible in cell tracking.
One of Ultrack’s most remarkable traits lies in its methodological innovation. Traditional cell-tracking algorithms typically separate the workflow into two independent steps: first segmenting the cells in each frame of a time-lapse video, then associating or linking these segmented cells between frames to construct their trajectories. This sequential approach struggles with ambiguities inherent in microscopy data, such as distinguishing whether a large blurred region corresponds to a single cell or multiple overlapping cells, or discerning the aftermath of cellular division events. Ultrack redefines this paradigm by jointly solving segmentation and linking in a unified framework. This dual adjustment process enables the algorithm to dynamically refine cell boundaries considering their behavior and continuity across sequential frames, thus dramatically improving accuracy.
Ultrack’s core computational engine revolves around the creation of an ultrametric contour map—a hierarchical representation of potential cell boundaries extending from coarse outlines to finer partitions within each frame. By evaluating which segmentations maintain the highest consistency over time and across neighboring frames, Ultrack identifies the contour configurations that best represent real cells in a biologically plausible manner. This approach mimics human perceptual strategies; for example, just as our brains infer whether a passing cloud is a single mass or a pair of smaller clouds by observing their continuity and relative motion in the sky, Ultrack infers cell boundaries through temporal coherence.
Further refining its predictions, Ultrack integrates fundamental biological constraints into its model. It inherently understands that while cells can divide into two, they very rarely merge or transform abruptly between distant locations. These rules help the software eliminate implausible segmentations or tracking jumps, substantially reducing false positives and tracking errors. Consequently, Ultrack not only accelerates computational processing—saving hours typically spent on correcting mistakes—but also empowers researchers by minimizing manual intervention, particularly in dense tissues where manual corrections can be prohibitively labor-intensive.
An additional major advantage of Ultrack is its generalizability across datasets. Unlike many tracking tools that require retraining deep-learning models for each new dataset—a process that is time-consuming, data-intensive, and computationally demanding—Ultrack delivers accurate tracking without such retraining. This flexibility significantly lowers barriers for labs wishing to adopt the technology, allowing immediate application to a diversity of biological systems.
To validate Ultrack’s performance, the research team selected the zebrafish neuromast as a model system. The neuromast, a mechanosensory organ vital for fish navigation, offers a well-characterized but challenging setting for cell tracking across developmental time. Following the standardized guidelines of the Cell Tracking Challenge, Ultrack achieved near-perfect accuracy in segmenting and tracking these cells. This exceptional performance underscores the tool’s robustness in handling real biological complexities.
Ultrack’s developmental utility extends beyond embryonic models. Leveraging Ultrack’s capabilities, the Royer lab constructed Zebrahub, a comprehensive zebrafish cell atlas published in the journal Cell. This atlas reconstructs entire developmental trajectories, enabling a systems-level view of zebrafish embryogenesis. Meanwhile, other researchers at the Chan Zuckerberg Biohub have employed Ultrack to investigate immune system cells in zebrafish, highlighting its broad applicability across different biological questions and organ systems.
Recognizing the need to visualize and interact with the vast datasets generated by Ultrack, the team developed inTRACKtive, an innovative browser-based tool that allows users to explore cell trajectories in 3D space interactively. With functionalities such as rotating embryos, selecting specific groups of cells for detailed study, manipulating playback speeds, and even reversing time to observe developmental processes backward, inTRACKtive enhances the accessibility and interpretability of complex cell-tracking data. This interface was developed collaboratively between the Biohub and the Chan Zuckerberg Initiative, with significant contributions from scientist Teun Huijben.
To further extend the reach of their technology and datasets, Royer’s group integrated datasets from five additional model organisms—including the mouse, the nematode Caenorhabditis elegans, and the tunicate sea squirt—into a communal platform dubbed the Virtual Embryo Zoo. By harnessing inTRACKtive’s intuitive interface, scientists and educators worldwide can now explore embryonic development datasets interactively through any standard web browser, from desktop computers to smartphones. This open and user-friendly resource is designed to foster collaboration and data sharing within the developmental biology community.
The researchers encourage the broader scientific community to contribute their whole-embryo datasets to the Virtual Embryo Zoo. An expanding repository of embryonic developmental data across diverse species has the potential to revolutionize comparative developmental biology, facilitating insights into conserved and divergent cellular behaviors during organogenesis. Looking ahead, plans are underway to enhance inTRACKtive by integrating live imaging data directly with cell tracking results, thereby enabling richer multimodal visualizations that combine cellular movement patterns with contextual tissue morphology and function.
Ultrack represents a transformative advance in cell tracking technology by combining innovative algorithmic design, biological insight, and user-friendly visualization tools. It bridges a critical gap between the rapidly increasing availability of live imaging data and the analytical methods required to translate such data into meaningful biological knowledge. With growing adoption, tools like Ultrack and inTRACKtive are set to revolutionize our understanding of developmental processes, disease progression, and tissue dynamics, making intricately detailed cellular movies accessible to a broad range of researchers and ultimately accelerating discoveries across the life sciences.
Subject of Research: Cell Tracking, Developmental Biology, Computational Biology, Imaging Technologies
Article Title: Ultrack: pushing the limits of cell tracking across biological scales
News Publication Date: August 25, 2025
Web References:
- Ultrack paper: https://www.nature.com/articles/s41592-025-02778-0
- inTRACKtive paper: https://www.nature.com/articles/s41592-025-02777-1
- Zebrahub project: https://www.czbiohub.org/life-science/zebrahub-tracks-zebrafish-development/
- Virtual Embryo Zoo: https://virtual-embryo-zoo.sf.czbiohub.org/
References:
- Royer, L. et al. Ultrack: pushing the limits of cell tracking across biological scales. Nature Methods (2025). DOI: 10.1038/s41592-025-02778-0
Image Credits: Dale Ramos, Chan Zuckerberg Initiative (CZI)
Keywords: Cell biology, Imaging, Microscopy, Developmental biology, Computational biology
