Hi Nidul,

I think you mean the intensity of the immune reactivity or the overall score is 1, 2 & 3?

next step of scoring is to see the cells in an individual manner, like how many are 1+, 2+ & 3+ respectively. Get the proportions and you can make a score on your own.

What is the protein that you look for? There are some IHCs with established scorings already.

Cheers,

Ivan

Hi Nadul,

I would suggest a color deconvolution (depending on your staining http://www.mecourse.com/landinig/software/cdeconv/cdeconv.html ) and subsequent mearuement of the stained area after binarisation with ImageJ.

Best Soenke

Long answer below, but this is a more complicated issue than it would seem. There are several software approaches that exist to quantify IHC signal. Image J is the simplest, but it requires you to know the software, and know what you are doing. Some approaches are a bit more automated, like the Aperio scanner and analysis. These require expensive equipment, and from experience, despite the certain degree of automation, it can be difficult to have reliable results, as these depend on your staining to be very well done, free of artifact, and be very constant from slide to slide.

Also, there are theoretical issues with trying to quantify IHC staining intensity, the main one being that the relationship of the peroxidase - chromogen reaction (DAB or other) with concentration of the marker does not occur in a linear fashion; ie. if you have an area that is staining twice as intense as another area, that does not mean that it contains twice as much protein. So the results you obtain with quantification of IHC staining will always be relative. Using immunofluorescence is usually the correct approach when you want to have quantitative results.

It sounds from your question like you want to validate by image analysis the results that you obtained from your pathologist. You should be aware that any image analysis software approach requires in its development to itself be validated by a pathologist, to check that the numbers obtained by the software actually match the pathologist's scores. So in a way you're in a situation where you want to validate a pathologist's evaluation using a software approach that itself is validated by a pathologist... It's all a bit circular. If what you want is make sure that the pathologist's results are valid, what is usually done is to have the slides examined again by a second pathologist, blinded as to the first examination's results, and see how the results match. 

The method that we routinely use with brain tissue is to scan on a flat bed scanner at high resolution (ca. 4,000 dpi is fine); store the image as a JPEG, and open in Photoshop. The image can be minimally (and uniformly) modified using levels and the sharp filter, converted to gray scale and inverted; an ROI is selected using the lasso tool and the mean density determined from the histogram. Simple statistical methods can be used to compare groups of individuals.

There are some requirements for this to work, but most importantly the IHC staining cannot be saturated but must be proportional to the amount of epitope under consideration. This is best achieved by limiting the concentration of primary antibody and keeping the development time constant. We use nickel-enhanced DAB which tends to give higher contrast that regular DAB, e.g. a white background in gray-scale images.

In the past we have embedded various dilutions of  tissue homogenates, stained for particular epitopes and scanned stained sections. These show a complex non-linear relationship between concentration and density, but one can establish the range over which the response is proportional to the antigen concentration.

It is worth the effort to optimize for your particular antibody and the statictics can be convincing.

Good luck.

We use the Aperio positive pixel for all clinical trials.  It is a FDA approved algorithm.  You must first have the slides scanned via the Aperio system.  Then download the image scope  app to open the slides.   Free....The positive pixel algorithm is there.

It is the easiest  image analysis system there is.  Easier than  Image  J, Metamorph, and Image Pro plus.

You could also make use of "IHC Profiler" which was developed by our lab to quantify your IHC images. In addition to providing a score based on the protein expression, it will also accurately tell you the exact percentage of the number of pixels with varying positivity/score patterns.

It is a freely available plugin for ImageJ.

Link to the Paper: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0096801

Link to the IHC Profiler Plugin: https://sourceforge.net/projects/ihcprofiler/

Good luck. :)