Image: 3D (x,y,t) rendering of single-cell tracks belonging to a competitive MDCK epithelium.
This repository contains all the steps necessary for the annotation and tracking of a homogenous or heterogenous population of epithelial cells in time lapse microscopy data.
Each step has an associated Jupyter notebook which should be run in the following sequential order:
- Image alignment (align.ipynb)
- Cellular segmentation (cellpose_segmentation.ipynb/stardist_segmentation.ipynb)
- Phenotype classification (cellx_classify.ipynb)
- Object tracking (btrack_tracking.ipynb)
- Viewer (napari_viewer.ipynb)
The repository is designed to work with the raw output of any timelapse microscopy data set that is structured with the following path pattern:
/expt_ID/PosID/img_channel000_position000_time000000000_z000.tif/
i.e. Nathan/ND0000/Pos0/Pos0_aligned/img_channel000_position...z000.tif
This pipeline was originally designed to label a 2-class, mixed population cell competition images (a wild-type cell versus a mutant), but now has been modified to be more broadly applicable to a homogenous population of cell types, with the 2-class pipeline in a separate subdirectory.
Image loading is done via the DaskOctopusLiteLoader function from https://github.com/lowe-lab-ucl/octopuslite-reader