I have been trying to optimise chromogenic double staining for IHC. The staining it self seems to work quite nicely but I am having some problems with the quantification. Does anyone know a method or Image J plugin that could be useful? From the example figure I can use colour thresholds to separate green and red that can be converted to grayscale and quantified. However, in the middle of red dots you can see a small areas where the colours overlap. Is there any means to get information about red and green colour intensities from these overlapping sections separately for quantification.
With fluorescent this would be easier but here I can not just shut down the other channel or can I?
You can quantify the redness and greenness using the following formulas starting from the RGB color representation.
M=max(R,G,B)
m=min(R,G,B)
S=2(M-m)/(1+|M-0.5|+|m-0.5|)
r=atan2[ (G-B)/sqrt(2), (2R-G-B)/sqrt(6) ]
redness=S*max[ (cos(r)-cos(t))/(1-cos(t)) , 0 ]
g=atan2[ (R-B)/sqrt(2), (2G-R-B)/sqrt)6) ]
greenness= S*max[ (cos(g)-cos(t))/(1-cos(t)) , 0 ]
where 0
Hi, there is a well-working ImageJ plug-in for 'Color deconvolution' based on a method by Ruifrok:
http://imagejdocu.tudor.lu/doku.php?id=plugin:color:colour_deconvolution:start
Files are attached; I briefly tried your image and it looks promising. You could deconvolute the image measure the intensity of the separated channels.
Nice staining by the may, what are you looking at?
Best regards
PS: Customized vectors could yield a more specific separation, here is another try of mine. The value are best optimized with single-stained slides.
The image is quite bluish. I used for img1 u=pi/12 and for img2 u=pi/6;
Hi Vasile,
Thank you for your efforts, the images you have managed to get look good! What software are you using? Could I just change color balance with tresholds to get rid of blue already before analysis, or the image could be imaged with more brighter ligh or use even lighter hematoxyllin treatment. I am afraid that my mathematics and programming skills are too inadequote to take use of the formula that you provided earlier. But after seeing your results with it I will definately try to find some help with that.
Andreas,
Thank you Andreas also for the imageJ link. I have played with its settings for a while now and the results are getting better. The problem is still to get low backround but without losing colour intensity values to get something similar that Vasile has managed to get. We use Dako permanent red and Biocare vina green for the staining. The figure is is from human breast section. Ultimately we would aim to stain and quantify a commercial breast tumor tissue microarray (TMA).
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Physical Phenomena
- Luminescence
- Fluorescence