Cellprofiler export to spreadsheet

Few quick questions:

  1. Where does the per-image mean, median, and SD values for object measurements output to? I can’t find them.
  2. Is there any way to automatically export the stat block produced when running MeasureObjects?
  3. When doing object math, is there anyway to ensure that the objects being divided are parent and child? It seems to go by number at the moment and this can cause problems.
1 Like

Hi @mdl54,

  1. Have you got three spreadsheets after analysing your images? The values you are asking about will be in your “Image” spreadsheet rather than your “punctafilter” one as they are an image level measurement rather than object level.

  2. Not that I’m aware of but hopefully the answer to the above question means it’s not necessary? It will mean your image spreadsheet might end up with many columns which can be annoying to navigate if you don’t use some sort of data analysis software downsteam. To ease this you can select which measurements you actually want to export or not in the ExportToSpreadsheet module.

  3. I’m not sure about this question really. Could you explain what you are trying to accomplish and the problems you are encountering?

Good luck,

Thanks for the reply,
You are correct, the answer to 1. does solve 2. As for question 3, I’m measuring vacuole intensity and plasma membrane intensity. I need to make sure that the measured vacuole intensity is divided by the plasma membrane intensity of the cell that it is in.

Hopefully I’m not leading you down the wrong path here but you would indeed want to run ObjectMath on objects that had already been paired properly (either by one compartment being created based on another and thus linked inherently by the same number, or using the Relate module if you are wanting to connect objects that somehow were not already related properly).

I’m not sure what @mdl54 wants to do is possible within CellProfiler and would have to be done post. I may be wrong in two different ways: 1. interpreting the question and 2. not knowing CellProfiler can do this.

I THINK what we want here is to divide each child measurement by the one measurement the parent they share has. But once you relate objects, the children are labelled by the parents object number so there’s no object that exists where it has it’s own child measurement and it’s own individual label and somehow its parent label as well?

I searched in the forum and found some other people discussing a similar question (below). Although the question being a bit older makes me concerned that this feature has been added and I’m just failing to make it work when I try.

1 Like

That works, thanks. Follow up question, is there a way I relate objects based on proximity? I only need to measure the plasma membrane of the cell and this creates an object that has a hole in the middle, meaning the relate objects module doesn’t work as is. Alternatively, if there was a way to only measure the outer rim of an object that would work too.

The MeasureObjectIntensity module indeed produces metrics of intensity just for the edge pixels of an object, does that help?

I’m doing this from memory, not being in front of the software, but if Relate does not allow you to define relationships by proximity instead of overlapping, then one solution is to expand objects using the ExpandOrShrink module (I think that’s what it’s called?) until they DO overlap.