-2

I have binary segmentation masks of nuclei edges which I want to analyze post segmentation. Specifically, each object needs to be identified individually. Many of the objects are overlapping or sharing an edge making it difficult to automatically separately identify these objects with different labels. The only supporting image beside these labels would be the original H&E images.

Is there anyway to separately identify overlapping edge objects?

Sample Image Example

Ive been attempting to use connected components with OpenCV library or a distance transform. The issue is this doesn't help with overlap. Matlab or python solutions would be best!

D Aslam
  • 1
  • What have you tried so far? SO is no programming service where you put in a question and get out some code. I would start with skeltonizing the image. Then I would try to seperate the different overlapping objects by analyzing junctions. But that is very opinion based. Be prepared that the community will close your question, because it is too broad. – Markus Aug 30 '23 at 05:37

0 Answers0