Any programmatic tools apart from image J to quantify microscopic images from histological staining?

Question

[charlesgwellem]

I am very new to image analysis. I wish to find out if any one can help me with information about programmatic tools of any type that can be used to quantify microscopic images from histological stainings.

I have performed a Sirius red staining to stain for collagen in my tissues. With image J, I discover that the collagen content is exagerated by normal areas of high collagen staining (which also stain red) such as the areas of the airways, compared to pathological areas with high collagen content. this exageration is partly because I find it very difficult, almost impossible, to set a colour threshold to clearly distinguish between normally high collagen-content areas and pathologically high collagen-content areas. I want to be flexible enough to prevent this exageration from normal high collagen-content areas.

Thank you in advance for your kind help.

[Joe]

I don't think you'll stand to gain much from moving to a different tool (especially compared to the venerable Fiji/ImageJ), as an image is an image at the end of the day.

That said, for python, you can look at simplecv/opencv2 which is a very powerful computer vision package, as well as scikit-image. Both have tools for analysis of image data (the latter being particularly popular amongst AI practitioners). You can also get quite a lot done with numpy and matplotlib. All of these solutions are going to require to you get your hands seriously dirty with code though - I'm not sure how you feel about that.

For bash, you can also do far more with imagemagick than people realise (see here for instance). I've never explored computer vision in R, so someone else will have to weigh in for that one!

If you are struggling with discerning features in your images, I would suggest you look at what image manipulations you can apply upstream which might help (edge detection, binary filtering, morphological operations etc etc. Without seeing the images its hard to know what might help).