Texture Granularity analysis

Hi all!

I am new to Cell Profiler and I am processing a bunch of images containing 1 object only. The “texture” inside the object changes with time, from one main intensity point in the centre of the particle, to several appearing with in time. I have the interest to measure the number and size of these “intensity hot-sports”. I am exploring the use of “Measure Granularity” “MeasureObjectIntensity” and “MeasureObjectIntensityDistribution” but have several questions.

-Firstly, my particles are dark on white background so I invert the images as first step before identifying primary objects. It is not clear for me if the three mentioned objects require the “intensity hot-spots” in white (my original grey scale image) or if they also work as dark on white background (the inverted image). Since I am only interested in analysing what is happening inside the object, it is not clear if the measure object intensity and distribution work correctly in the original grayscale image since the object is detected on the inverted image.

-Measure granularity: I have set subsampling factor for granularity measurements 0.25; subsampling factor for background reduction 0.25; radius of structuring element 10; range of the granular spectrum 20. I have taken radius if structuring element as a mean value, assuming with the range it screens lower and larger radius sizes. As a result in my spread sheet I get 12 columns, and I don’t know how to interpret them, since I am running a small test with 5 image containing 1 to 4 “intensity hot-sports”.

-“MeasureObjectIntensityDistribution” I get would be the right module, if it would automatically detect the intensity peaks and measure the distribution from them, but I understand from the help that it is only taking one center.

Is there any adjustment I can make, or am I looking into the wrong modules? I would much appreciate your advice!

texture_1png texture_3

It sounds like what you actually want to do is to identify the “hot spots” as objects (using a second IdentifyPrimaryObjects module); you can then measure their size and intensity (with MeasureObjectSizeShape and MeasureObjectIntensity), and then optionally use RelateObjects to relate them back to the larger main object if that’s helpful.

To briefly answer your other questions-

  1. You can use MeasureObjectIntensity on either your pre-inversion or inverted image, either is fine.
  2. MeasureGranularity is going to measure whatever spectrum of sizes you tell it to, it isn’t looking for smaller sub-objects, thus the number of columns will always be the same no matter how many “hot spots” are present, it’s only the VALUES that will change. I think the diagram in the module help might help you understand specifically what those measurements refer to.
  3. Correct, it only takes one center.
1 Like

Thanks @bcimini!
I’ll give that a try then