Segmenting epithelial tissues

I am trying to segment epithelial tissues, like the tif image attached, and track them in time. I have attached my pipeline that does it reasonably well. However, if you run it you can see that it tends to oversegment. I have played around with watershedding and things in the past, so I am guessing this is due to shallow minima, but I am not sure how to get rid of it here and segment more accurately.

Any suggestions?

Mani_Forumquestion_pipeline.cp (7.97 KB)

Hi Madhav,

I’m attaching a pipeline which does a better job, though it could probably still use some tweaking. I should warn that I’m guessing that object tracking will probably be tricky with images like this. i.e., little separation between cells.

The approach you used is good; the only substantive changes are the following:

  • Smoother background subtraction in CorrectIlluminationCalculate

  • Filtering method and size in Smooth.

  • Changed thresholding method in IdentifyPrimaryObjects

  • Probably most important, manually adjusted the smoothing filter size in IdentifyPrimaryObjects.

2011_09_23.cp (6.41 KB)