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.
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 .
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