How to adjust Area and Perimeter into um from pixels

Hello,

I am new to Cell profiler and am trying to get the collected perimeter and area data in um, not pixels. Any advice on how to do this?

Also, my pipeline often segments the cells where segmentation should not occur, leading to a higher cell count. Is the only solution editing my images prior to running the program?
Attached are the images of the bad segmentation & area/perimeter output. I’ve also attached the pipeline I’ve haphazardly created, which isn’t exactly working.

Thanks in advance for any support!

cellprofilerqa|690x449 .

cellprofileqadata

bb_pipeline4.0.cpproj (455.4 KB)

Hi @jjwpark,

To convert to um you’ll need to find out the scale factor at which your particular images were captured, since the microscope used will effect how big a single pixel is in microns. This can often be found in the image metadata.

The segmentation you showed looks like it’s been set up to try to separate touching objects too aggressively. You might want to try altering the expected object size parameters, alongside the method for distinguishing clumped objects. Unfortunately it’s difficult for us to test your pipeline without a sample image to run with it.

Hope that helps

@DStirling Thank you so much for your response!
I’ve attached a sample image, if that helps.
The scale bar is attached.
cellprofilertest

Also, do you know how I can set theIdentifySecondaryObjects module to recognize the cells within the cell wall? I’ve photoshopped the images prior and that seems to work but I’m wondering if there is a quicker method to accomplish this.

Thanks again.

Hi @jjwpark

Is the example you provided one of the actual images that you’re loading into the pipeline? In general it’s best not to be editing images before trying to perform analysis, and usually you’ll need to export an image without a scale bar in order to analyse properly.

Could you also confirm what you mean by “cells within the cell wall”? Is this the only channel that you captured or are there other stains involved?

Thanks