Shape of primary object


#1

Hi there

I am segmenting nuclei as primary objects. I am wondering which would be the easiest and fastest way to discard weirdly shaped primary objects (1).

Also cellprofiler sometimes makes 2 primary objects out of one(2). If I increase the pixel size I will get more of the weirdly shaped objects tho. I guess there is no way to avoid this, or is there?

thank you!



#2

Hello,
It can be a challenge to find the sweet spot. What you can try is changing the method used to distinguish clumped objects and method to draw dividing lines between clumped objects (e.g. intensity vs shape).
You can also try to pass the image through MedianFilter or GaussianFilter prior to segmentation. Another option is resizing the image by Resize module prior to analysis. This can also help remove this over-segmentation issue.
Finally, you can filter (remove) the segmented weirdly shaped objects based on specified criteria. This can also be done when analyzing data in the produced spreadsheet.