DynaLink is a state-of-the-art tracker that predicts consistent cell trajectories robust to occlusions, shape changes and shifts in motion. Point it at a live-cell acquisition and it returns tracks that hold one identity per cell from the first frame to the last, with everything it needs already in the box.
One acquisition, played through. Every detected cell carries an identity and draws a trail behind it as the run proceeds. A red cross marks each point where DynaLink relinks a trajectory across a gap, so the same identity continues on the far side of it.
Single live-cell acquisition, playback accelerated.
Immunotherapy readouts are posed over long observation windows, and tracking is asked to survive conditions there that exceed what a single motion model can describe.
T cells elongate and deform as they move and engage, so appearance-based association drifts over the course of an acquisition.
Cells pass in front of one another and their targets, leaving gaps that break a trajectory in two.
Movement during scanning differs from movement during attack. A model fitted to one state fails on the other.
An identity switch part-way through divides one cell's history into two. Any event defined at the level of the individual cell is quietly lost.
Biomedical cell motion behaves much like a stochastic differential equation. We simulate SDEs at volume — including the appearance of the occlusion gaps themselves — and apply the resulting model to real acquisitions.
Given two interrupted tracks separated by a gap, DynaLink estimates the likelihood that they belong to the same cell, and links them where that likelihood is high.
Starting from scratch, DynaLink is the whole thing: give it an acquisition, get long-horizon trajectories back. If you have already validated a tracker and a detector, it works just as well on top of them, extending the tracks you already trust.
A companion tool exposes merge, split and stitch operations on trajectories, for curation, quality control and ground-truth generation.
A T cell attacks an organoid and kills it. If the trajectory breaks at the wrong moment, the analysis records a miss. The error surfaces at the level of interpretation, which is where it is expensive.
A T cell may kill several cancer cells in succession, hopping between them. That behaviour is only observable if identity is preserved across the hops.
A newer and increasingly relevant readout, and one that resolves only over long observation with identity preserved throughout.
Advancing a compound requires specific evidence in the preclinical package. Long-horizon readouts that hold up let treatment validity be tested convincingly.
DynaLink sits after standard live-cell imaging workflows and runs on your existing acquisition, instruments and protocols. Reliable automated long-term readouts mean more lead candidates can be evaluated with the setup you have.
We would be glad to run DynaLink on one of your own long acquisitions and show you the comparison side by side.
Talk to us