Sirius red/fast green staining

Image Analysis
#1

Hello Lovely People,

I started using the current CP for quantification of collagen area(red area) in hearts and kidney tissue, but I have some difficulties and most the pipelines in the forum does not seem to work in the current version 3.1.8.
Below are sample images (unfortunately cannot be uploaded). Can someone please help me out with a suitable pipeline to build on in a step-wise approach.
Much appreciated

Analyzing Picrosirius Red polarized light images
#2

Please find the link to the images below,

https://drive.google.com/drive/folders/1jsc0CSZwVsdr5Gw8Q9Ti5PxqPrXhduVM

#3

Hi there!
You probably would want to use three steps (after loading your images into CellProfiler using the first 4 default modules, Images, NamesAndTypes, etc):

  1. UnmixColors to isolate the red regions. ColorToGray may also do the trick if it’s a nicely separable stain.
  2. Threshold to decide what counts as stained vs not stained. There are many options for this module.
  3. MeasureImageAreaOccupied
    Give it a try!
#4

Hi Amponsah

Continuing on with @AnneCarpenter.

  1. UnmixColor extract the fast red stain
    22%20PM

  2. Identify what is not tissue and mask those pixels

    40%20PM
    40%20PM548Ă—560 125 KB

  3. Then identify what is red stained. Change threshold correction value to be more or less stringent with the OTSU automatic threshold
    12%20PM

  4. Measure Image Area Occupied. Collagen = 9.9% of total tissue
    55%20PM

Cellprofiler v3.5.1 Collagen.cpproj (646.8 KB)

Best
Lee

2 Likes
#5

Thanks Anna much appreciated

#6

Much appreciated Lee but a quick follow up: with regards to the identifyprimaryobject module;

What do you consider/identify as Not tissue and how do you determine its diameter in pixels and also same for the diameter of red stain since they are spread over the tissue.

Many thanks

#7

Hi Amponash

This very much depends on your images. If you follow the pipeline you can see that anything with a diameter from 200pixels to (the entire size of the image) allows you to identify most of the non-tissue areas. I would keep a minimum size because you don’t want to remove very small non-tissue regions as they may be close to objects you wish to count, especially if you are using the “remove if touching edge”. For the diameter of the red stain you can see I did not put a size restriction as there is no need for the question you are asking. This means that even single pixels that pass your threshold will be identified as collagen. This is because you want to know the proportion of tissue that is collagen.

I hope this helps.
Best
Lee

#8

Many thanks @lee.marshall and sorry for my late reply.
Does that mean the Green stain is what you consider as the non tissue areas?

Also from the second identifyprimaryobject module (collagen/red stain) the diameter was set between 10-40, so how do you mean u did not put a size restriction on it? Is it because of not discarding objects outside this range?
Lastly is possible to get in touch with you outside this forum, will really appreciate some guidance.
Much appreciated for your assistance. Have a lovely weekend.

#9

Hi Amponsah

Green output is all the collagen in the sample, all the white in the original image (non tissue pixels) were masked so they don’t count to the total pixels. Yes choosing the “discarding objects outside this range” ignores the size of the objects.

Send me a personal message and we can discuss further

Best
Lee