Cytoplasmic immunostain intensity

[Ana Yuil-Valdes]

Dear all:

QuPath is excellent!

I am new in the tissue analyzing software world. I am using whole slide images. I am having troubles with the intensities. I will show a case that I know is 3+ (high). I am using the positive cell detection option.

Then I used the automatic intensity parameters 0.2, 0.4 and 0.6 for cytoplasm DAB OD mean. This means  that:

•  less that 0.2 is 1+ (weak)
• 0.2-0.4: 2+ (medium)
• more than 0.6: 3+ (high)

My results:

I got high number of 1+ and 2+ and few 3+?

What do I need to change?

My guess is that I need to set up my own threshold. Maybe 0.02 (1+), .0.05 (2+) and 0.1 (3+)

Another question for the annotation result table I got the number of 1+, 2+ and 3+ for each polygon. Can I get the total for all the polygons. For example if I analyze 10 polygons I would like to see the total number of 3+ cells in 10 polygons.

Thank you for your help and support,

Ana

[micros...@gmail.com]

The easiest way to handle your last question would be to select all of your annotations before you perform the cell segmentation and Objects -> Merge Selected Annotations.  

If you need both separate values and summary values, that is more of a scripting problem, which is somewhat covered in these two places:

https://gist.github.com/Svidro/8f9c06e2c8bcae214cdd7aa9afe57c50#file-using-selectionmodel-for-groups-of-objects-groovy

https://groups.google.com/forum/#!topic/qupath-users/gr1jHeFSmJc

You would need a combination of the two.

The initial values for your stain intensities are just filler, but should fit many staining profiles.  It is expected that you will adjust those as staining can vary from batch to batch, antibody to antibody, tissue to tissue.

Also note that you are looking at Cytoplasmic OD mean, which may or may not be what you actually want.  Use the H key or the transparency slider to see if it might be more accurate to go off of the "nuclear" or "whole cell" signal.

[Pete]

As an alternative to merging annotations (which might be slow), you can also just create a new annotation that completely contains all 10 of your existing polygons.  By 'completely' I mean that they can't overlap (even a tiny bit!).

The easiest way to do this might be to simply create an annotation around the entire image with Objects -> Create full image annotation (shortcut Ctrl + Shift + A).

That way the summary measurements for the large annotation should include everything 'below' it.  There's an illustration at https://github.com/qupath/qupath/wiki/Object-hierarchies#summary-statistics

(You can ignore the fact that TMA cores are shown there, it's the same principle when you have annotations fully inside other annotations.)

Apart from that, I agree that you might want to score using a different measurement (either nucleus or cell, rather than cytoplasm) in addition to adjusting the thresholds.  I wrote a bit about how I'd approach that at https://petebankhead.github.io/qupath/tips/2018/03/22/setting-positive.html

[micros...@gmail.com]

Oh, nice.  All this time and I had never noticed that the detection summaries merged for larger annotations!  Always more to learn.

[Ana Yuil-Valdes]

Thank you! I merged the annotations using "Merged Selected Annotations" Thanks microscopyra

I changed my threshold parameters. Is working!! 

QuPath is amazing. Thanks Pete

Thanks

Ana