Whole slide image (WSI) analysis

Dear CellProfiler Team,

I am trying to use the CellProfiler 2.0 to analyse a section from a whole slide image. As you can see in the attached image DefaultOUT__3.mat (77.9 KB) by zooming in on the objects of interest (purple stained RNA) some of the detail is lost.

Additionally, I also tried to run the attached pipeline DefaultOUT__3.mat (77.9 KB) which is a modified version of one of the pipelines suggested in a previous post ‘Quantifying colours in histology’ in the forum for CellProfiler 1.0.

The issues that I need help with are:

  1. Is there a way of pre-preparing the image in order to enhance its quality for use with CellProfiler 2.0? Do you have any suggestions for WSI?
  2. Is it possible to use the pipeline to calculate the ratio of the RNA molecules (purple) to the Nuclei (grayish stained).

Thank you for your help.

Hi Fatima,

Zooming in will not alter the resolution that CellProfiler has available, i.e. no detail will be lost. There is a interpolation of the pixels for display only, so that when you zoom in it looks like you are seeing a fuzzy image, when in fact you are just resolving individual pixels and we have interpolated between them. (We have since taken this “feature” out and in the next release only the true pixel values will be displayed.) Am I guessing correctly what you are reporting?

For histological slides, I would point you to our UnmixColors module:
This might help you ID particular stains better than your current ColorToGray module (not sure, but give it a try!).

I made a quick re-write of your pipeline. Check it out and see if I am even on the right track, or if it helps at all. I made some notes in the Comments field at the top.

Re: (1) As for “pre-preparing” the image, it depends on what the goal is. A good place to start is EnhanceOrSuppressFeatures module.
Re: (2) Do you mean the count of RNA vs. the count of Cells? Or the mean intensity of the ratio of all the RNA molecules vs. the Cells?

Let us know if this helps!
Pipeline_DLogan.cp (8.88 KB)

Hi David,

Thank you for addressing all of my issues. Your pipeline was on the right track and your comments guided me as well. The only question that remains is that can CellProfiler be used to analyse whole slide images (WSI)? The image that I had shared earlier was a static image of a section of the WSI. We would ideally like to analyse the WSI in one go.

Thank you for your help,

Hi Fatima,

Glad to help. Regarding WSI images, I see now – I didn’t realize before that this was a particular image format. I thought you were just referring to large images of whole slides generically (sorry, we don’t typically work with these sorts of images).

CellProfiler uses Bioformats to offload the problem of reading the plethora of image formats in the ever-expanding imaging universe. However looking at this list I don’t see .wsi files listed. More than likely WSI files are some form of TIFF files, so what happens when you try to load them in CP?

Would you upload a sample WSI file for us to take a look at? If we can’t figure it out, we do know the Bioformats folk in Wisconsin, so maybe we can persuade them to write a new method to parse WSI files :smile: