Quantifying Necrosis in Liver H&E slides

[Austin Moore]

Hi all!

Apologies if this is a silly question, but I'm very new to using these sorts of programs.  I am trying to use QuPath to quantify the percentage of tissue that is necrotic on H&E slides.  The first steps seem simple enough--I use stain vectors and tissue detection to mark out my tissue (giving me the total area of the tissue).  I then use SLIC superpixel creation to divide into tiles, use Compute Intensity Features on the tiles to add features, and then classify into necrotic and non-necrotic tissue.  Looks great, if I look at the map it is divided well into the groups I want.

The silly part that I am stuck on--how do I pull total area from each of these classes so that I know the percentage of total area?  I have the number of tiles that are necrotic and non-necrotic, but all of the features detailed don't include the area.

Is there an easy way to compute the area from this step?  Or should I be going about this sort of classification a completely different way?  Thanks for the help!

[micros...@gmail.com]

Sounds basically like this, though you can look around the other scripts to fine tune it.

https://gist.github.com/Svidro/68dd668af64ad91b2f76022015dd8a45#file-area-measurements-to-annotation-groovy

Though you would want to check for class and add two different measurements. I will try to create a more generalized and useful version in a little while.

[micros...@gmail.com]

All done.

https://gist.github.com/Svidro/68dd668af64ad91b2f76022015dd8a45#file-class-and-area-measurements-to-parent-annotation-groovy