Program for tracking cell nuclei in (intestinal) organoids over time. Uses a convolutional neural network for nucleus detection, a min-cost flow solver (Haubold, 2016) for linking nuclei over time and tools for manual error correction.
- Manual tracking with live error-detection.
- Automated cell detection using a convolutional neural network based on U-net.
- Support for automatically linking nuclei over time using a min-cost flow solver (Haubold, 2016)
- Supports TIFF files, TIFF series, Leica LIF files and NIKON nd2 files.
- Plugin API with live-reload for fast data analysis
- Do some manual tracking to obtain ground truth data and training data.
- Train a neural network.
- Apply the automated tracker on some new time lapse movie.
- Correct the errors in the tracking data of that time lapse movie.
- Use the corrected tracking data as additional training data for the neural network.
- Want to track another time lapse movie? Go back to step 3.
Tutorial on manual tracking
Tutorial on automated tracking
OrganoidTracker must be installed using Anaconda. See the installation page for details. If you are updating from an older version of OrganoidTracker, its dependencies might have changed, and in that case you also need to visit the installation page.
Open an Anaconda Prompt, activate the correct environment and navigate to the
The organoid_tracker.py
script starts a graphical program that allows you to visualize and edit your data.
After you have installed the software, please have a look at the manual. The manual is also available from the Help
menu in the program; this works even when you're offline.
- The original published network - for images with low background noise, 0.32 μm/px, z-step 2 μm
- For lower resolution - for images with low background noise, 0.41 μm/px, z-step 2 μm
You can also use OrganoidTracker as a library to write your own scripts. All public functions in OrganoidTracker have docstrings to explain what they are doing. As a starting point for using the API, see the API page.
It's possible to use OrganoidTracker from Jupyter Notebooks. Just install the notebook
conda package into your OrganoidTracker environment and everything should be ready. Detailed instructions to get you started are available at the Jupyter manual page.
Install the program as normal, and then point your Python editor (PyCharm or Visual Studio Code are recommended) to this directory. Make sure to select the organoid_tracker
Anaconda environment as the Python environment.
The files dealing with the neural network are licensed under the MIT license. This is indicated at the top of those files. Other files are licensed under the GPL license. Please cite the publication in PLOS ONE if you're using this work.