Improvement of 3d segmentation for the cells

I am the first time user to cell profiler. I want to do 3d segment of the attached image.
I followed the pipeline from here but not getting the correct output.


After watershed algorithm it is telling me that my segmented image has 3,35,140 cells. This is wrong because each cell get segments in multiple pieces not as a whole.
This is screenshot output of Rescale filter and watershed filter.

I am sharing the a small piece or original tif data and the pipeline what I made to do the segmentation.
If anyone know how to improve the pipeline please help me. Thank you.
Here is the link of input data and the pipeline

The attached pipeline doesn’t seem to be reading/loading your tif image properly. What you would want is to have the frames separated. Then assign the names to channels in the NamesAndTypes module based on a column that is different between the channels, could be ImageName or Channel, or something that you extracted from the file name.

Also, there is no nuclei image attached, but in general:
I think you could use a filter (Median or Gaussian for example, depend on your image and the segmentation goal) to smooth your image.

If your segmented region gets break down, you could try increasing the Downsample and Footprint in the Watershed module.

If you want to use the Nuclei objects as seeds for the cells, in Watershed, you could use Markers.
You can also do all the segmentation in one pipeline.

Hope this helps!

Thanks a lot Nasim. I don’t know why it did not work. Sorry for that. The thing is that I have only cells no nuclei. Is it possible to do proper segmentation of cell membrane without nuclei. Like median or Gaussian filter you said that is for nuclei not for cell Correct? My main question is how to do proper water segmentation of cells without using nuclei objects as seeds.
Thanks a lot.

I don’t think you necessary need to have a seed object for your segmentation, proper use of filters, thresholding strategy and watershed could give you the segmentation you’re looking for. However, I think the challenge in your images is that the intensity of the edges are higher than the middle, so you may need to try several things to make it work.