Hi Michael and Brad,
What you are asking is basically a colocalization question, which is a rather touchy subject.
The correctness, accuracy and usefulness of your question will seriously depend on what your images are and how they were acquired.
1. Brightfield images with a color camera.
I know that there has been success if the stainings are easy to separate (red-blue (HRP-AP), Turquoise-Red (b-gal-AP) and Red-Blue (Seq double AP) )
...You can then use Fiji's Color Deconvolution Tool, to extract the two color channels and then measure the ratios of pixels positive for your stainings by applying a threshold for example. I could show you how if you email me.
Other kinds of stainings I have little experience, though I would be careful about the results. I know a company called Nuance has a spectral Imaging camera that can theoretically separate triple classical IHC stanings... untested)
2. Fluorescence images with a color camera (Or B&W camera)
Here you have quite the literature about it, just google "Fiji Colocalization Analysis" for a tutorial on it. For you, since I don't know what your images are, there are three options:
a. You have an RGB image but there is only Red and Green that have some signal. Then you can just run "split channels" (CTRL + L) in Fiji if you need to find where the command is and then the Colocalization module
b. You have a multichannel image already, then it's easy, you just need to use Split Channels again and then measure the colocalization
c. You have an RGB image that has a bit of everything. Then I would be careful in the processing and would suggest you post an image so we can assess the feasibility of your request.
Best
Oli
Olivier Burri
Engineer - Image Processing
& Software Development
EPFL - SV - PTECH - PTBIOP
-----Original Message-----
From: ImageJ Interest Group [mailto:
[hidden email]] On Behalf Of Michael Schmid
Sent: vendredi 25 janvier 2013 10:49
To:
[hidden email]
Subject: Re: Percent of Red Pixels in Green Pixels
Hi Brad,
there are different ways to select the red pixels in an RGB image, e.g. Adjust>Color Threshold, which is a simplified version of Gabriel Landini's Threshold Color. Select a range without the ends with the Hue sliders and unclick 'Pass', and select roughly the upper half of Saturation (with 'Pass' on). When you like the result, click 'Select'.
You can also try the Versatile Wand plugin: e.g. have a 'typical' red color as eyedropper (foreground) color, select connectedness=non-contiguous and 'eyedropper'.
http://imagejdocu.tudor.lu/doku.php?id=plugin:segmentation:versatile_wand:startThen use Analyze>Measure to measure the area selected, and compare it to the full image area. (Analyze>Set Measurements must have the 'Area' checkbox selected).
One more possibility, though less flexible: convert the image to a HSB stack and use Adjust>Threshold on the Hue channel.
Michael
________________________________________________________________
On Jan 25, 2013, at 03:43, 余明俊 wrote:
> Hi,
> I am trying to count the percentage of green pixels that are also red. Is there a method for this? I understand that I will have to decide what is red and what is green. Is there a way to get the pixel values using ImageJ. Thank you very much.
> Brad
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
