Guide dense cell segmentation in phase contrast using fluorescence image information

I am working on a project to track HeLa cells and gain information on the cell cycle. The cells have a Fucci reporter system, which gives a fluorescent signal in the nucleus of cells, green or red depending on the phase of the cell cycle. The segmentation and tracking of cells is based on the corresponding phase contrast images as the cell divisions are clearest in those types of images.

The problem I have is that the cells are very densely packed, which makes segmentation hard in the phase contrast images and gives lots of undersegmented groups of cells. It is also hard to see by eye where one cell stops and another one starts. I was wondering if there is any way in Ilastik to use the information in the fluorescent images (i.e. the stained nuclei) to aid in the segmentation. What would be very useful is that Ilastik would segment the cells in such a way that each ‘object’ automatically contains only one nucleus to avoid undersegmentation.
Does anyone know of such a solution or how to approach this?

Many thanks!

1 Like

Hi @Arno,

welcome to the Forum! :slight_smile:

from what it sounds, you have a promising setup going. While I think I know what can be done there, as I am not a biologist, I am not 100% sure what your images look like. Would it be possible for you to maybe share an overlay of all the channels you have available?