Tumor tissue - tumor vs stromal


I just started using this program and I think it is very powerful. However, when trying to identify the primary objects (nuclei stain), I cannot only select cells that are tumors. In other words, cells that do not have elongated nuclei/cytoplasm. The only option I have is the minimum pixel and max pixel lengths. I need some way to reject nuclei that have, say, a major axis/minor axis ratio of 2 or 3. Is there a way to do this? If not, is there another way to do it?

I would prefer not doing post-processing via the excel output because I cannot see what is actually being excluded. Thanks!

You might you want to try something like the following:

  • Identify nuclei as you have already.
  • Use MeasureObjectSizeShape to make per-cell morphological characteristics.
  • You can use DisplayDataOnImage or ClassifyObjects to pick a measurement of choice and display the value of the measurement on an image of the nuclei. That way, you can eyeball what value you’d like as the cutoff for the measurement. In your case, it sounds like Eccentricity might useful. Or you can use CalculateMath to computer the aspect ratio from the MajorAxisLength and MinorAxisLength measurements, and use that.
  • Once you’ve decided on a measurement and cutoff, use the FilterObjects module to filter objects that exceed that cutoff. Keep in mind that the new objects created by this method will need to have measurements made, i.e, they do not inherit the measurements of the objects they came from.

Hope this helps!

Thanks a bunch! I will try that today

Should I use “Rules” and use a ratio calculation in the text file? Do you have an example of how this might be done in terms of what the text file would contain? Thanks!

nevrmind. I found the “math” on the pulldown list to choose my ratio.

I figured out how to filter out the long nuclei or cells. However, is there a way to use the nuclei eccentricity as the filter but filter out the corresponding cells? I basically have primary objects, filter out eccentricity>0.9, find secondary objects (cells) based on nuclei

That didn’t work so well. I’ve also tried to filter using eccentricity of the actual cells but that is not as accurate. Thanks!

To filter out the Secondary objects, just place IDSecondary after the FilterObjects module, using the filtered object name as input.

Hope that helps,