Segmenting spores and ignoring haploids



I’m working on a project with a collaborator right now, and she’d like to use CP to identify S. pombe spores in an image that features both spores (4 nuclei in one cell) and haploid cells (1 nuclei in 1 cell).

She’s integrated both a nuclear and a faint cytosolic marker in the RFP channel. I’m wondering if there’s a way that I can use CP to identify the nuclei (RFP), and then identify spores in the same image by segmenting the cells around the nuclei based on the fainter RFP signal, creating cell outlines that encompass either 1 or 4 nuclei.

Is that possible? So far I’ve only been able to incorrectly segment spores into 4 cells.

Sorry for the poor description, it’s late and it’s Friday!



Hi Erin,

Is your collaborator intending to use CP 2.0 or 1.0? They will have a much better shot at using 2.0 based on your description.

In CP 2.0, you can using Otsu global in IdentifyPrimaryObjects in a 3-class mode which biases the threshold towards the foreground or background pixels, and the user gets to chose which. They can then take the following steps:

  • Use Otsu global with the middle class set to foreground to identify the cell perimeters based on the dim RFP pixels
  • MaskImage to mask the image with the cells.
  • Per-object Otsu with the default two class or with the middle class set to background to identify the nuclei within the cells
  • Use RelateObjects to establish the relationship between the nuclei and enclosing objects.
  • ClassifyObjects to classify the cells on the basis of the number of children (i.e., nuclei) that they have. Or FilterObjects canbe used similarly to filter out the type that is unwanted.




Thanks Mark, I’ll suggest that! I think they were planning on using CP1, but if 2 is easier switching shouldn’t be a problem.