Calculation of basic features like intensity, area etc


Hi Team,

Im using CP for sometime now and really felt very thankful to all of you for such a great software.
I have some basic queries to be clarified:

*]How exactly, Integrated intensity is calculated?
Is it that once primary & secondary objects are identified, then we simply take the pixels falling into each object and then adding the intensity of each of these pixels together to get the integrated intensity. For example, nucleus.

*]I’m also wondering, how the area and perimeter of cells are calculated. Are any geometrical shapes considered for area calculation, especially irregular shapes like cytoplasm?

*]I guess, Mean intensity is simply = integrated intensity/area??

*]How can we convert area into number of pixels and volume, if needed?

*]Whats the size of the pixel considered during analysis? Is is fixed or varies from each set of images?

*] Since, nucleus is having nucleolus (mostly multiple), is there anyway by which we can remove the area occupied by nucleolus so that we can get the actual Mean intensity by using refined area. This is because, many a times nucleolus cant be stained with dyes like DAPI but while calculating we consider whole area against only stained parts by dye as integrated intensity?

Thanks a lot for making me understanding all these issues.
Mridul KK



Integrated intensity is the sum of the pixel intensities within an object- if you are only considering the nucleus (primary object) the integrated intensity is just the sum of the pixel intensities within this object; a secondary object (such as a cell) includes the nucleus, so integrated intensity would be the sum of the pixel intensities within the nucleus and cytoplasm.

The area is computed from the actual number of pixels in the region; the perimeter is the total number of pixels around the boundary of each region in the image (so there is no ‘fitting’ of any geometrical shape for an irregular object- this is captured more in measurements like solidity and extent).

Yes, mean intensity is the average pixel intensity within an object (integrated / area).

If you go to File->Set Preferences, that’s where you can enter the pixel size in microns (thus it does not vary from image set to image set unless you change it). All measurements are reported in pixels, so area is in fact = # pixels.

As for your last question, this depends on how you identify your nucleoli. If they are stained with an entirely separate stain from your nucleus, then cropping them out of the DAPI image will allow you to see only the nucleus. If all you have is the DAPI, and the nucleoli stain more brightly, using ApplyThreshold to threshold away the bright nucleoli and then measure only the rest of the nucleus should work.

hope this helps!



Hi Kate,

Thanks for really nicely explained answers.

Just couple of queries:

  1. How can we calculate Volume from # of pixels, if possible?

  2. If I have two cells and looking for translocation of a protein from cytoplasm to nucleus, whats the best way to determine: in which cell, the translocation has taken place more among the two?
    Should we use integrated intensity or mean intensity and in either case, which ratio will be good?

  3. I guess, sum of integrated intensity of nucleus and cytoplasm should be equal to integrated intensity of whole cell. Then why, the integrated intensity of cell is lower than the sum of nuc+cyto intensities. This is what, we have observed in our data. The difference is b/w 20-100.

  4. we have observed that for unstimulated cells, whose cytoplasm has high intensity (~high protein conc), their respective nucleus is also showing high intensity though visually, there is no or very low protein is present in nucleus. But for cells, whose cytoplasm has low intensity, the nucleus intensity is pretty low, correlating well with what is visually observed. Does, high cytoplasmic intensity impacts the measurement of nuclear intensities?

I hope, the answers to all these questions will actually place me in a very understandable stage in terms of concept.

Thanks for bearing so many queries.
Mridul KK



To answer some of your questions:

(1) You are asking to obtain a three-dimensional measure (volume) from two-dimensional quantity (pixel). You would need to know the depth of field of your microscope to approximate this, which is something you would have to find out from the microscope vendor or literature.

(2) Assuming the amount of protein remains constant within a particular cell, I would think that the cell with the higher proportion of nuclear protein to total protein would be the one that has experienced more translocation.

(3) It would be helpful to upload an actual image set (not the CellProfiler figure window, as Kate mentioned elsewhere). We would need an image from each of the channels and then we could use the pipeline that you uploaded here, if it is the relevant one.

(4) This is possible, by virtue of the fact that a bright object (cytoplasm) surrounding a dim object (nucleus) may increase the apparent brighness of the dim object, especially if the depth of field is large. Using a lower exposure time while imaging or using a confocal microscope can help with this.