|
INTERNAL SEMINAR: 1) Escaping the bottleneck: collective neutrophil migration in complex environments; 2) From raw data to automated segmentation and tracking (with minimal effort)
Par Theresa Jakuszeit, Irshad Abibouraguimane
Le 17 Mars 2026 à 11h00 - TBA
|
Résumé
Theresa Jakuszeit:
Neutrophils are not only the first responders to infection, but also among the fastest migrating cells in the human body. As they migrate toward sites of inflammation, they navigate the complex tissue architecture by squeezing through pores far smaller than their resting size. While the behaviour of individual migrating cells has been studied extensively, their collective dynamics have received much less attention.
In this talk, I will present a previously unreported collective phenotype in which neutrophils form traveling bands of high cell density that move up a chemical gradient. I will then show how these dense cellular waves pass through bottlenecks, exhibiting qualitatively different behaviour depending on the geometry of the confinement. These results suggest that neutrophil behaviour at bottlenecks differs significantly from passive or non-deformable active systems.
Irshad Abibouraguimane:
Segmentation and tracking of moving objects in videos is a challenging problem.
Training neural network for this task requires annotated data that takes a lot of effort and time to produce.
I will present a workflow to train models from scratch. The workflow allows us to go from raw data to annotated data
with minimal effort using a human-in-the-loop approach. I will show how this workflow is fully implemented
in the ImageJ plugin BACMMAN and demonstrate how we use it to train our segmentation and tracking model Distnet2D on images of bacteria.
