Segmentation with high overlap

Hi, I’m currently trying to automate the segmentation of yeast cell images. I’ve tried Weka Segmentation but the lack of clear boundary and amount of overlap between cells seems to be getting in the way. Is there a better way to go about this? Thanks!

CF1.tif (6.0 MB)


your cells are quite variable and dense, so not so trivial. You can try introducing a third class “border” in Weka segmentation. Trace some cell boundaries for this class, it can help separating cells.

If you happen to have any other fluorescent channel, combining channels mathematically is also often useful.