Line scan analyses of individual cell border intensities?


#1

Hi Folks,

I am looking to accurately measure the intensity of cell borders, on a per-cell basis, of human endothelial cells that will either exist in pseudo monolayers or in isolation depending on the treatment. My strategy thus far has been to:

  1. Identify primary objects (nuclei) by DAPI staining (IdentifyPrimaryObjects module)
  2. Identify secondary objects (cells) by HCS CellMask staining (IdentifySecondaryObjects module)
  3. Expand the “cell” objects by a specified number of pixels (ExpandOrShrinkObjects module)
  4. Identify tertiary objects (putative cell borders) by selecting the larger identified objects as “expanded cells” and the smaller identified objects as “cells” (IdentifyTertiaryObjects module)
  5. Measure the intensity of the “cell borders” (MeasureObjectIntensity module)

Although this strategy is somewhat effective, I have realized that it is limited by several factors including:

  1. The cell-cell variability in how well the morphology of the identified cell objects matches the morphology of the actual cell borders.

  2. The cell-cell variability in how close the identified cell objects are to the actual cell borders.

  3. The cell-cell variability in how thick the actual cell borders are.

To circumvent these issues, I was thinking that line scans of a user-defined length could be created perpendicular to the identified cell object’s edge in a radial manner in order to measure the integrated intensity of the cell borders on a per-cell basis (see cartoon below).

Does anyone know if these sorts of analyses or anything similar would be possible to do in CellProfiler? Any thoughts or suggestions would be greatly appreciated!

Many thanks!

KG


RadialCV explanation
#2

This is a good idea, though I think that MeasureObjectRadialDistribution (recently renamed to MeasureObjectIntensityDistribution) would do what you want. The lines drawn are orthogonal to what you have drawn, i.e. they are annuli, but if you had enough annuli and they tracked the border well enough, you would be satisfied.

There is a discussion here, on the (now updated) module in questin: https://github.com/CellProfiler/CellProfiler/issues/1255
E.g. the visualization looks something like this now:

There is no built-in way to measure the intensities in CP currently in the way you have drwn them, i.e. in polar angles around various radii from the center of the object (though I have thought this a useful addition!).

NOTE: You need to use the Beta here to get this visualization in MeasureObjectIntensityDistribution.


#3

Hi David,

Thanks for the quick response and for passing along the info about the MeasureObjectIntensityDistribution module. I will download the beta version of CP and give the module a shot.

Since you mentioned that the aforementioned line scan analyses could be a useful addition to CP, do you know if this is something that might be developed in the near future or if this is something that could potentially be developed via the CP support plan?

Many thanks for your assistance and I will report back soon!

Best,
Kyle


#4

Hi Kyle,
Yes, it may be possible to do this with our support plan, but it depends on our group’s scientific and software engineering priorities, too, i.e. we don’t take pure work-for-hire jobs. But if it aligns with our needs/wants, then it’s possible. I’ll follow up with you via email.

David