deepcell-tracking uses deep learning models from deepcell-tf within an assignment problem framework to track cells through time-lapse sequences and build cell lineages. The assignment problem is solved using the Hungarian algorithm.

Getting Started

deepcell-tracking is a Python package that can be installed with pip:

pip install deepcell-tracking

Or it can be installed from source:

git clone https://github.com/vanvalenlab/deepcell-tracking.git

cd deepcell-tracking

# install the dependencies
pip install .

How to Use

from deepcell_tracking import CellTracker

# X and y are the time-sequence data and their corresponding segmentations (labels), respectively.
# model is a deepcell-tf tracking model.
tracker = CellTracker(X, y, model)

tracker.track_cells()  # runs in place, builds tracks

# Save all tracked data and lineage files to a .trk file
tracker.dump('./results.trk')

# Open the track file
from deepcell_tracking.utils import load_trks

data = load_trks('./results.trk')

lineage = data['lineages']  # linage information
X = data['X']  # raw X data
y = data['y']  # tracked y data