I’m trying to figure out whether to use the old colocalization threshold or Coloc2 or JaCoP for measuring the number of pixels, area percent or area where an immunofluorescent protein in one channel colocalizes with a cellular structure or a cell type marker in another channel. This just needs to be in a 2-dimensional dual immunofluorescent image. The lower limits of the threshold for each channel are initially set with the negative control slides (non-immune serum) and the exposure and gain on the camera used for the images for each channel are set so that saturation is not reached. The lower limit of the threshold is then confirmed with slides (or regions) where no staining is reasonable (e.g. the erythrocytes of a control would not normally have a nuclear marker). We then want to see how much of the channel A area is located in channel B. Intensity is really not the issue if it exceeds the threshold (provided the controls worked). We want to see which cells and organelles express the marker of interest; e.g. what percent of channel B is located in the same place as channel A and what percent of channel B is in A and what is the total area of co-localized A and B/total area (of length of a structure) has been evaluated. Can this be done using Coloc2 or JaCoP and, if so, how?
Bump. I’d like to know the same. I’m trying to look at spinal cord tissue sections and want to see the colocalization of markers and want to get percent overlap between the different markers.
Similar. I’m looking at filament fusion where a mix of red and green filaments become yellow upon fusion (sharing cytoplasm). My approach has been to binarise the images for object based colo. However, I can’t do object-based coloc since these aren’t spots (which seems to be a prerequeiste for such analyses).
Now I’m using JACOP to report the Pearson’s correlation between my binarised green and red filaments.
I would like feedback on this approach, thanks.
This is a follow-up. We can use the old colocalization plugin and are able to accurately do this. By converting both the green and the red fluorochrome images to 8-bit grey scale and thresholding the selected image (you will need threshold each of the images separately) and then use co-localization (do NOT select co-localized points 8 bit). In the new co-localized image, you then convert to an 8 bit greyscale. You can then measure the co-localized points in the region of interest using a threshold of 255. You can copy the region of interest from the co-localized image using ctrl c (copy) followed by ctrl v (paste) on one of the original thresholded images (green or red) and then ctrl z (removes the copied tissue leaving only the outline) and move the region over the same tissue on the first fluorochrome and repeat for the second. The area of each fluorochrome above the set threshold can then be measured as well as the area of co-localization. We tried to do this with co-localization finder but we seem to only get the outline of the co-localized areas. We have also been unable to do this with JACoP and Coloc2 but these seem to be intensity based. In tissue sections we have low level background but distinct positive fluorescence above some pretty clear threshold levels but the intensity is going to vary - JACoP and Coloc2 do not seem to be able to provide the analyses we need. Has anyone tried one of the other programs for this? Is there a way to do this using JACoP or Coloc 2 or Co-localization finder or a different plugin? Any chance that Colocalization will continue to be supported for those doing area-based image analysis of multiple markers fluorescence in formalin-fixed tissue sections?
I solved this issue in the end by scripting the image calculator to process the thresheld R + G images. I calculated the total pixles (OR) and the co-loc pixels (AND) and expressed as % coloc within the sample.
I was even able to do this in 3D (with non-overlapping sections) by turning my stacks into montages and running the calculator on them.
Let me know if you need more details.
Did you do that using JACOP?
No, I ran my own processing (de-noising, binarisation etc.) then used the image calculator for OR and AND images, with the area being measured for each.
I simply wanted to know the pixel/pixel % overlap, where i couldnt find a simple solution so made my own. It certainly depends on your application with regards to binarisation or if you want to retain the pixel values.