Cell SCT

Create a machine learning model that can Segment, Classify and Track (SCT) cells.

This is a three part project, where the end goal is to have a model that can, from a series of micrographs,

• segment out the cells; 'seg'
• classify them by their cell cycle phase; 'ccc'
• and track them over time.

To track the daily progress of this project: journal

Initial setup

• Start with the environment_setup Python Notebook. It will install the correct packages into your environment and make sure the GPU is connected.
• There is also an omero setup notebook to establish a connection if needed.
• Right now, the segmentation part of the project is done and can be used to train a cell segmentation model and evaluate it qualitatively and quantitatively with comparison options.
• The classification part is under works.

What works

• The environment and omero connection setup notebooks are done.
• The cell segmentation model training notebook is done. It can be used by anyone to understand, with little code to understand, how to train a CellPose2 model.
• The cell segmentation model evaluation notebook is done and presents the functions to qualitatively and quantitatively evaluate a model and compare different ones.
• The omero setup notebook is done and can be used to collect data as well as save data.
• The training data notebook is done and can create a training data CSV file from 4 channel images and apply the classification rules on them.

What I'm working on

• Classifying the cells from the PCNA channel only.