Visual Tracking of Cell Boundaries and Geometries

Research Project Goal

Professor Tomlin's group in the Department of Aeronautics and Astronautics and Professor Axelrod's group in the Department of Pathology, Stanford School of Medicine, are collaborating on a number of projects aimed at constructing mathematical models to understand cell signaling networks and developmental phenomena in Drosophila melanogaster (fruit fly). In particular, the following are two topics under current investigation: the study of the feedback mechanism underpinning planar cell polarity (PCP) in the Drosophila wing; and the study of segmental groove formation in the fly embryo. In the first project, a mathematical model is developed which describes the directional polarization of epithelial cells. This polarization, which is reflected in the hair formation on the surface of the wing, is due to signaling events and cellular responses (feedback mechanisms). The second project looks at how the segmental groove cells are designated, what signals direct their change in shape to drive groove formation, and how they execute this response. In both instances, the cells in the sheet have irregular shape, change their shape, move, and divide throughout time. The developed mathematical models are tested on rich data sets coming from numerous experiments performed in the lab. These data sets mostly come in the form of colored, high-resolution images or videos. The research goal for this project is to design and implement a visual tracking algorithm that follows the movements and shape modifications of the mentioned networks of epithelial and endothelial cells. Extracting this information from the data provided by the experiments performed in Axelrod's laboratory would allow its effective use in the building and testing of the corresponding mathematical models.

Research Scope

It will be key to come up with a tracking method capable of recognizing cell boundaries and vertices, assigning 'tags' to single cells, and of following their relative movements. The scheme should be able to cope with noisy and over-abundant data. As mentioned above, the project may allow two levels of implementation: a tracking algorithm for the first study, which is a planar 2D case, and the extension of the tracking algorithm to work on the data for the second study, which is instead a volumetric 3D case.

Sample Data

• cell shape movies

Tasks

• To be further defined. However, roughly, the implementation steps may be the following:
• Data acquisition and conditioning;
• Noise Filtering;
• Tagging;
• Tracking;
• Storing the outputs in a ready-to-use format.

Challenges

Research Project Status

student names here

Point of Contact

Claire Tomlin, tomlin@; Jeff Axelrod, axelrod@; Alessandro Abate, aabate@stanford.edu

Midterm Report

not yet submitted

Final Report

not yet submitted