24 Aug 2020

2D Cell Tracking Example

A 2D cell tracking example from a multichannel time series data set. Contains a segmentation recipe, modules needed to track the cells through time, and a video of the finished result.


The images are matched 2D phase contrast microscopy and EFFP labeled fluorescent microscopy across 61 time points. The fluorescent signal comes from a cell cycle regulator called cyclin-dependent protein kinase regulator (Clb2). Raw data are found here: http://cellimagelibrary.org/images/35769. Tips:

  • You must use Amira for Cell Biology to load and replicate this project.
  • Toggle Auto-Refresh if you want to use the Process Time Series module.
  • The Time module governs all of the time sliders throughout the project and this port is pinned.
  • The pixel size is 0.2134 um x 0.2134 um.
  • The project was created with Spatial Units disabled. This is recommended if there are any errors when loading the data with Spatial Units enabled.
  • The tracks display slightly offset from the images. If you can’t see the tracks, try moving the camera to the other side of the 2D image.
  • You can recreate the video with the animation director.
  • The Recipe Player may need to be provided with a corrected recipe path in order to work on your system. Be sure that the recipe path is "{SCRIPTDIR}/BuddingYeastSegmentation_2June2020-files/BuddingYeastSegmentation_2June2020.hxrecipe" and replace "
    {SCRIPTDIR}" with the path to where you loaded the project from.

Recipe steps:

  1. Load fluorescent and phase contrast channels.
  2. Run Non-local means filter on phase contrast channel.
  3. Run gaussian filter on output of step 2.
  4. Extract and subtract background image from output of step 3.
  5. Use Variance module on the output of step 4 to find slow-changing bright background to make segmentation easier.
  6. Use arithmetic to perform thresholding on the output of step 5.
  7. Morphological operators on the output of step 6 to remove small particles and holes. Also inverts this result. This will be a reverse mask for later.
  8. Change input to fluorescent channel and run non-local means filter.
  9. Adaptive Thresholding on output of step 8 to select high intensity regions.
  10. Morphological operators on the output of step 9 to remove small spots, remove small holes, and separate cells using Separate Objects module.
  11. Labeling and addition of the output of step 7 using an arithmetic module.
  12. Marker based Watershed Inside Mask module using output of step 11 as markers, output of step 10 as a mask, and the image data to provide a gradient image.
  13. Finally, the background (from step 7) is removed using arithmetic and the output is a label field of separated and labeled cells.

David Ball, Jean Peccoud (2011) CIL:35769, Saccharomyces cerevisiae. CIL. Dataset. https://doi.org/doi:10.7295/W9CIL35769